oncogenes and prognosis in childhood t-cell acute ... · oncogenes and prognosis in childhood...
Post on 16-Mar-2020
0 Views
Preview:
TRANSCRIPT
Oncogenes and prognosis in childhood
T-cell acute lymphoblastic leukaemia
by Nicholas G Gottardo MB ChB, FRACP (Paeds.)
10268591
A collection of papers presented for the Degree of Doctor of Philosophy to
the School of Paediatrics and Child Health, University of Western
Australia
March 2008
Principal supervisor: Adjunct Prof Ursula R Kees PhD Telethon Institute for Child Health Research, Perth Co-supervisor: Dr David L Baker MBBS, FRACP (Paeds.), FRACPath Formerly of Department of Haematology & Oncology, Princess Margaret Hospital for Children, Perth Secondary supervisor: Assoc Prof David Forbes MBBS FRACP (Paeds.) School of Paediatrics, University of Western Australia, Perth
i
Preface
The regulations of the University of Western Australia provide the option for
candidates for the Degree of Doctor of Philosophy to present their thesis as a series of
papers which have been published in refereed journals, manuscripts that have been
submitted for publication but not yet accepted, or manuscripts that could be submitted.
All manuscripts presented relate to the study of genes involved in the prognosis of
childhood T-cell acute lymphoblastic leukaemia (T-ALL). The papers have been
presented in a logical format so as to address the issues raised in the introduction. The
discussion integrates the papers and puts the work in the context of current clinical
practice and how the findings may improve future prognostication and therapy for
childhood T-ALL. Each paper is presented with the original internal headings, figures
and tables however, for ease of reading and flow, the formatting in the thesis is
uniform. Headings, figures and tables are numbered in the thesis style.
All of studies presented were of the candidate’s own design in consultation with the
candidate’s supervisors. All studies were performed in Adjunct Prof Ursula R Kees’s
division, within the Telethon Institute for Child Health Research, Perth, Western
Australia. The contribution of others is presented below and also at the end of each
chapter. Supervisors and co-authors assisted with the corrections and proof reading of
manuscripts following completion of drafts.
ii
Publications arising from this thesis (first and joint first*author)
Gottardo NG, Jacoby PA, Sather HN, Reaman GH, Baker DL, Kees UR. (2005)
Significance of HOX11L2/TLX3 expression in children with T-cell acute
lymphoblastic leukemia treated on Children's Cancer Group protocols. Leukemia.
Sep;19(9):1705-8. (Chapter 2)
Author contributions: NGG designed and performed the research, analyzed the data
and wrote the manuscript. PAJ carried out the statistical analysis. HNS provided
statistical data and advice. GHR was involved in the concept for the study and he
contributed many of the specimens and clinical data. URK and DLB supervised all
aspects of the study and preparation of the manuscript.
Dallas PB, Gottardo NG*, Firth MJ, Beesley AH, Hoffmann K, Terry PA, Freitas JR,
Boag JM, Cummings AJ, Kees UR. (2005) Gene expression levels assessed by
oligonucleotide microarray analysis and quantitative real-time RT-PCR -- how well do
they correlate? BMC Genomics. Apr 27;6(1):59. (Chapter 3)
Author contributions: PBD and NGG contributed equally to this work and were
responsible for designing the study, analysing, collating, and interpreting the data, and
preparing the manuscript. MJF carried out the statistical analysis, AHB and KF assisted
with data analysis, experimental design, and data interpretation. PAT, JRF, JMB, AJC
and NGG carried out the microarray and qRT-PCR experiments. URK supervised all
aspects of the study and preparation of the manuscript.
iii
Gottardo NG, Hoffmann K, Beesley AH, Freitas JR, Firth MJ, Perera KU, de Klerk
NH, Baker DL, Kees UR. (2007) Identification of novel molecular prognostic markers
for paediatric T-cell acute lymphoblastic leukaemia. Br J Haematol. May;137(4):319-
28. (Chapter 4)
Author contributions: NGG and URK conceived the study. NGG was responsible for
analysing, collating, and interpreting the data, carrying out the qRT-PCR experiments
and preparing the manuscript. MJF and KP carried out the statistical analysis under the
supervision of NHdK. AHB and KH assisted with data analysis, experimental design,
and data interpretation. KH and JRF carried out the microarray experiments. DLB and
URK supervised all aspects of the study and preparation of the manuscript.
Gottardo NG, Ford J, Baker DL, Kees UR. The triternepoid CDDO enhances
doxorubicin mediated cytotoxicity. Manuscript in preparation (Chapter 5)
Author contributions: NGG and URK conceived the study. NGG was responsible for
carrying out all the experiments, analysing, collating, and interpreting the data, and
preparing the manuscript. JF provided the cell lines for the experiments and technical
assistance. URK and DLB supervised all aspects of the study and preparation of the
manuscript.
iv
Abstract
The treatment of childhood acute lymphoblastic leukaemia (ALL) is one of the great
success stories of paediatric oncology, transforming a universally fatal disease into one
where 75 to 90% of children are now cured. Although in the past survival for children
with T-cell ALL (T-ALL) lagged behind that of children with pre-B ALL, the use of
contemporary intensified treatment strategies has significantly diminished this
difference, with many investigators reporting similar cure rates for both groups of
patients. Despite these marked improvements, numerous challenges still face
physicians treating children with T-ALL. Firstly, there have been no additional major
improvements in outcome over the last decade, despite additional treatment
intensification. Secondly, effective regimens remain elusive for treating children with
relapsed T-ALL or patients with resistant disease. Finally, there is a need to identify
patients currently potentially overtreated and thus unnecessarily subjected to acute and
long term toxicities without benefit. A major challenge therefore, is the identification of
novel reliable prognostic markers, in order to identify patients at high risk of relapse
and conversely those least likely to relapse, to guide therapy appropriately. Children
predicted with a high risk of relapse would be candidates for intensification of therapy
and/or novel experimental agents. Conversely, patients predicted to be at low risk of
relapse could be offered clinical trials using reduced intensity therapy, thereby
minimising toxicity. The consensus among investigators is that advances in therapy are
unlikely to be made without increased knowledge of T-lymphoblast biology.
The primary aim of this study was to improve on traditional prognostic markers and
develop a predictive test to be used at the time of diagnosis to determine whether an
individual patient is likely to be successfully treated on current therapy or has a high
risk of relapse. A secondary aim was to exploit the genetic differences between
v
relapsing and non-relapsing T-ALL patients to facilitate the development of novel
therapeutic targets.
This study found that, in contrast to other reports, HOX11L2 overexpression was
associated with a favourable outcome. Indeed, no patient whose T-lymphoblasts
overexpressed HOX11L2 suffered a relapse. This finding has significant clinical
implications, since such heterogeneity in expression of possible molecular prognostic
markers potentially allows T-ALL to be stratified into risk groups that predict disease
behaviour more precisely.
In order to identify a molecular signature of relapse we used DNA oligonucleotide
gene expression arrays to characterise the gene expression profiles of an unselected
cohort of T-ALL patients treated on the former Children’s Cancer Group (CCG)
protocols, some of the patients later suffering a relapse. We identified a novel set of 3
genes (CFLAR, NOTCH2 and BTG3), termed 3-gene predictor, which distinguished
patients with a favourable outcome compared to patients with an adverse outcome.
Crucially, the 3-gene predictor was validated in a completely independent cohort of
T-ALL patients, also treated on CCG style therapy. Our 3-gene predictor appears to
identify a high risk group of patients which require alternative therapeutic strategies in
order to attain a cure.
This study has also identified a potential novel agent for the treatment of T-ALL, which
may be used as an anthracycline potentiator or anthracycline-sparing agent. We
hypothesised that genes associated with a relapse signature provide promising targets
for novel therapies. We tested the hypothesis that CFLAR, an inhibitor of the extrinsic
apoptotic pathway and a member of the 3-gene predictor may be involved in the
development of resistance to chemotherapy. To test our hypothesis we used a novel
agent, 2-cyano-3, 12-dioxooleana-1,9 (11)-dien-28-oic acid (CDDO), previously shown
vi
to inhibit CFLAR protein, in two cell lines established in our laboratory from paediatric
patients diagnosed with T-ALL. We found that CDDO displayed single agent activity
at sub-micromolar concentrations in both cell lines tested. Importantly, minimally
lethal doses of CDDO resulted in significant enhancement of doxorubicin mediated
cytotoxicity in one of the cell lines assessed.
The findings presented as part of this thesis have revealed the value of gene expression
analysis of childhood T-ALL for identifying novel prognostic markers. This study has
shown that expression profiles may provide better prognostic information than
currently available clinical variables. Additionally, genes that constitute a relapse
signature may provide rational targets for novel therapies, as demonstrated in this
study, which assessed a potential novel agent for the treatment of T-ALL.
vii
Acknowledgements
There are many people I would like to acknowledge and without whom this thesis
would not have been possible.
Firstly, I would like to thank my principal supervisor, Adjunct Professor Ursula Kees
and my co-supervisor Dr David Baker. I could not have had better mentors for my PhD
thesis. To adjunct Professor Ursula Kees, you have an incredible knowledge and insight
for the subject and a tremendous passion for science. Thank you for your continued
enthusiasm, encouragement and guidance during my research and your ability to
explain things clearly and logically. Also, thank you for teaching me how to design
experiments and present data in a logical manner. To Dr David Baker, you are quite
simply an inspirational paediatric oncologist. Thank you for all your support, advice
and mentoring during my PhD thesis and Paediatric Haematology/Oncology
Fellowship. I have learned an amazing amount from you. Also, thank you both
immensely for the constructive comments and suggestions during the preparation of
this thesis.
I would also like to thank my secondary supervisor Associate Professor David Forbes.
Thank you for our informal chats. Your level headed approach and wise advice
regarding professional and personal matters was greatly appreciated.
I am especially grateful to Philippa Terry, who diligently and patiently taught me most
of the laboratory techniques which I used to undertake this PhD thesis. I am also
indebted to Jette Ford for her technical advice and assistance and for providing a
constant supply of cell lines for my research.
I would also like to thank Marty Firth, Peter Jacoby and Kanchana Perera for all the
statistical analyses. Thank you for your patience and willingness to explain statistical
principles and approaches to me.
viii
Thank you to Alex Beesley, Katrin Hoffmann and Peter Dallas for your readiness to
share advice and for advancing my knowledge of molecular biology.
To all the members of the Division of Children’s Leukaemia and Cancer Research,
Telethon Institute for Child Health Research and Centre for Child Health Research,
University of Western Australia, including, Simone Egli, David Holthouse, Stewart
Cattach, Darcelle Dixon, Renae Weller, Richard Hopkins, Joseph Freitas, Aaron
Cummings, Nadia Milech, Misty-Lee Palmer, Joanne Boag and Reinette Orr. Thank
you all for your help and advice during my research and for having made the lab a
stimulating environment in which to learn and a real pleasure to work in.
To the patients and parents involved in this study, a special thank you for entrusting us
with your precious samples.
I would also like to thank the National Childhood Cancer Foundation Laura and Greg
Norman Fellowship for supporting my research.
Finally and most importantly, thank you to my wife Carolyn and daughter Alyssa. This
thesis would not have been possible, without your continued love, support,
encouragement and patience.
In closing I would like to dedicate this PhD thesis to Kyle Andrews, a special young
man whose incredible courage was only overshadowed by his enormous heart.
ix
Abbreviations
AIEOP Associazione Italiana Ematologia Oncologia Pediatrica AIF apoptosis inducing factors ALL acute lymphoblastic leukaemia AML acute myelogenous leukaemia Apaf-1 apoptotic protease activating factor 1 array-CGH comparative genomic hybridisation BFM Berlin-Frankfurt-Münster BM bone marrow BMT bone marrow transplantation CALGB Cancer and Leukaemia Group B CCG Children’s Cancer Group CCR continuous complete remission CD cluster of differentiation CDDO 2-cyano-3, 12-dioxooleana-1,9 (11)-dien-28-oic acid CLL chronic lymphocytic leukaemia CLP common lymphoid progenitor CML chronic myeloid leukaemia CNS central nervous system COG Children’s Oncology Group CRT cranial radiotherapy CSF cerebrospinal fluid DFCI Dana-Farber Cancer Institute DHFR dihydrofolate reductase DMSO dimethyl sulfoxide DISC death inducing signalling complex
x
DN double negative DOX doxorubicin DP double positive DR death receptors EFS event-free survival EORTC European Organisation for Research and Treatment of Cancer FACS fluorescence activated cell sorting FADD Fas-associated death domain FDA Food and Drug Administration FISH fluorescent in situ hybridisation FLT3 FMS-like tyrosine kinase 3 FPGS folylpolyglutamate synthetase GO gene ontology IT intrathecal LSC leukaemic stem cell MHC major histocompatibility complex MM mismatch MRCUK Medical Research Council United Kingdom Council MRD minimal residual disease MTXPG methotrexate polyglutamate NCI National Cancer Institute NHMRC National Health and Medical Research MTT 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide NOD/SCID Non-obese diabetic/ severe combined immunodeficient OS overall survival PCA principal component analysis
xi
PDGF-R platelet-derived growth factor receptors Pre-B B-precursor PM perfect match POG Paediatric Oncology Group PGR prednisone good response PPR prednisone poor response PPTP Paediatric Preclinical Testing Program q-PCR quantitative polymerase chain reaction qRT-PCR quantitative real time reverse transcriptase polymerase chain reaction RER rapid early responder RF random forest RFS relapse-free survival RMA robust multi-array analysis SCID severe combined immunodeficient SER slow early responder siRNA short interfering RNA SNP single nucleotide polymorphism SP single positive T-ALL T-cell acute lymhoblastic leukaemia TCR T-cell receptor T-NHL T-cell non-Hodgkin’s lymphoma TIT triple intrathecal therapy TNF tumour necrosis factor TRAIL TNF receptor apoptosis-inducing ligand WBC white blood cell
xii
List of Figures
Figure 1.1 T-lymphocyte development
Figure 1.2 A proposed schema for selecting the optimal therapy for children with T-ALL at presentation
Figure 2.1 Expression of HOX11L2/ACTB and HOX11/ACTB in 40 paediatric T-ALL specimens
Figure 2.2 Clinical outcomes for 39 paediatric T-ALL patients according to expression status of HOX11L2 and HOX11
Figure 3.1 Examples of Pearson's correlations between gene expression levels determined by qRT-PCR and oligonucleotide microarray
Figure 3.2 Pearson's correlations between fold-change in average gene expression levels between subsets of interest assessed by qRT-PCR
Figure 4.1 Unsupervised hierarchical clustering of the training T-ALL cohort using the 300 top Random Forest-ranked probe sets
Figure 4.2 Percentage representation according to Gene Ontology biological function category of 300 top-ranked probe sets
Figure 4.3 Cumulative incidence of relapse for the validation cohort (n = 25) of T-ALL patients stratified by the 3-gene predictor
Figure 4.4 Mean (SEM) expression levels for genes comprising 3-gene predictor by quantitative real-time reverse transcription polymerase chain reaction, for adverse and favourable outcome patients for combined cohorts
Figure 5.1 Level of CFLAR expression, as measured by Affymetrix microarray gene expression analysis
Figure 5.2 Cytotoxic effects of A) doxorubicin (DOX) alone and B) CDDO alone, on T-ALL cell lines PER-427 and PER-604
xiii
Figure 5.3 PER-427 and PER-604 cells were cultured with CDDO at IC50 concentrations for each cell line
Figure 5.4 PER-427 cells were incubated in the presence of CDDO at A) the minimal dose (0.195µM) and B) IC50 concentration (0.8µM)
Figure 5.5 Cytotoxic effects of DOX alone (open bars) or in combination with the minimal dose of CDDO (0.195µM) (grey bars)
xiv
List of Tables
Table 1.1 Results of selected clinical trials in T-ALL by comparison with pre-B ALL conducted over the last 20 years
Table 1.2 Frequency and prognostic significance of molecular abnormalities found in T-ALL
Table 3.1 A comparison of average qRT-PCR, RMA, and MAS 5.0 scores and the corresponding correlation values for the 31 transcript-concordant genes assayed in this study for which the Affymetrix microarray probesets (Affy IDs) were deemed likely to recognise identical transcripts to qRT-PCR probes
Table 3.2 A comparison of average qRT-PCR, RMA, and MAS 5.0 scores and the corresponding correlation values for the 17 genes assayed in this study for which the Affymetrix microarray probesets (Affy IDs) may not recognise the exact same transcript subsets recognized by qRT-PCR probes
Table 4.1 Patient characteristics
Table 4.2 Nine genes selected from the 300 top-ranked probe sets, which discriminated between adverse and favourable outcome
xv
Contents
Preface .............................................................................................................................i
Publications arising from this thesis ................................................................................ii
Abstract.......................................................................................................................... iv
Acknowledgements .......................................................................................................vii
Abbreviations................................................................................................................. ix
List of Figures ...............................................................................................................xii
List of Tables ...............................................................................................................xiv
CHAPTER 1. INTRODUCTION................................................................................1
1.1 Childhood T-cell acute lymphoblastic leukaemia ................................................1
1.2 Clinical features..................................................................................................4
1.3 Prognostic factors ...............................................................................................5
1.3.1 Cytogenetics ................................................................................................5
1.3.2 Immunophenotype and T-lymphoblast maturational stage............................5
1.3.3 In vivo response to therapy...........................................................................6
1.4 Treatment of childhood T-ALL...........................................................................8
1.4.1 Historical perspective...................................................................................8
1.4.2 Treatment phases .........................................................................................8
1.4.2.1 Induction...............................................................................................9
1.4.2.2 Consolidation........................................................................................9
1.4.2.3 Re-induction and interim maintenance ................................................10
1.4.2.4 CNS directed therapy ..........................................................................11
1.4.2.5 Continuation .......................................................................................15
1.4.3 Increasing dose intensity ............................................................................15
xvi
1.4.4 Chemotherapeutic agents........................................................................... 17
1.4.4.1 Anthracyclines and asparaginase......................................................... 17
1.4.4.2 Methotrexate....................................................................................... 17
1.4.4.3 Dexamethasone .................................................................................. 21
1.4.4.4 The anti-metabolite 6-thioguanine ...................................................... 21
1.5 Future strategies ............................................................................................... 22
1.6 Biology and new prognostic markers................................................................ 24
1.6.1 Origins of leukaemia – the cancer stem cell theory .................................... 24
1.6.2 The putative T-ALL cell of origin.............................................................. 26
1.6.2.1 Normal thymocyte development ......................................................... 26
1.6.2.2 T-cell development gone awry............................................................ 28
1.6.3 Oncogene activation in T-ALL .................................................................. 31
1.6.3.1 Homeobox genes: HOX11, HOX11L2 and HOXA-D cluster................ 33
1.6.3.1.1 HOX11 and HOX11L2 ................................................................. 33
1.6.3.1.2 HOXA-D cluster........................................................................... 34
1.6.3.2 Helix-loop-helix transcription factors: TAL1 (SCL), TAL2, LYL1,
bHLHB1......................................................................................................... 35
1.6.3.2.1 TAL1 (SCL), TAL2, bHLHB1 ....................................................... 35
1.6.3.2.2 LYL1 ............................................................................................ 37
1.6.3.3 LIM domain only zinc finger encoding genes: LMO1 and LMO2........ 37
1.6.3.4 MLL-ENL ........................................................................................... 38
1.6.3.5 CALM-AF10 ....................................................................................... 38
1.6.3.6 NUP214-ABL ..................................................................................... 39
1.6.3.7 MYB ................................................................................................... 40
1.6.3.8 NOTCH signalling in T-ALL .............................................................. 40
1.6.4 Tumour suppressor gene silencing in T-ALL............................................. 42
xvii
1.7 Future risk-stratification....................................................................................45
1.7.1 The genomic era ........................................................................................45
1.7.2 Identification of predictive markers............................................................48
1.8 Thesis hypothesis and objectives ......................................................................51
CHAPTER 2. SIGNIFICANCE OF HOX11L2/TLX3 EXPRESSION IN CHILDREN
WITH T-CELL ACUTE LYMPHOBLASTIC LEUKAEMIA ON CHILDREN’S
CANCER GROUP PROTOCOLS..............................................................................52
2.1 Abstract ............................................................................................................52
2.2 Introduction ......................................................................................................53
2.3 Methods............................................................................................................54
2.4 Results and Discussion .....................................................................................54
2.5 Author contributions.........................................................................................61
2.6 Acknowledgements ..........................................................................................61
CHAPTER 3. GENE EXPRESSION LEVELS ASSESSED BY
OLIGONUCLEOTIDE MICROARRAY ANALYSIS AND QUANTITATIVE REAL-
TIME RT-PCR – HOW WELL DO THEY CORRELATE? .......................................62
3.1 Abstract ............................................................................................................62
3.2 Background ......................................................................................................63
3.3 Results..............................................................................................................64
3.4 Discussion ........................................................................................................71
3.5 Conclusion .......................................................................................................75
3.6 Methods............................................................................................................75
3.6.1 Specimens..................................................................................................75
3.6.2 RNA extraction, preparation of target cRNA and hybridisation to
HG-U133A GeneChips.......................................................................................76
3.6.3 Processing and statistical analysis of microarray data.................................77
xviii
3.6.4 Bioinformatics........................................................................................... 77
3.6.5 qRT-PCR .................................................................................................. 78
3.7 Author contributions ........................................................................................ 79
3.8 Acknowledgements.......................................................................................... 80
CHAPTER 4. IDENTIFICATION OF NOVEL PROGNOSTIC MARKERS FOR
PAEDIATRIC T-CELL ACUTE LYMPHOBLASTIC LEUKAEMIA ...................... 81
4.1 Abstract ........................................................................................................... 81
4.2 Introduction ..................................................................................................... 82
4.3 Methods........................................................................................................... 83
4.3.1 Patient characteristics ................................................................................ 83
4.3.2 RNA extraction, preparation of target cRNA and hybridisation to
HG-U133A GeneChips ...................................................................................... 85
4.3.3 Quantitative real-time reverse transcription polymerase chain reaction ...... 85
4.3.4 Statistical analysis and bioinformatics ....................................................... 86
4.4 Results ............................................................................................................. 88
4.4.1 Gene expression profiles of newly diagnosed T-ALL patients ................... 88
4.4.2 Functional analysis .................................................................................... 90
4.4.3 Defining a gene expression signature predictive of outcome ...................... 92
4.4.4 Validation of the signature genes predictive of outcome ............................ 94
4.5 Discussion........................................................................................................ 97
4.7 Author contributions ...................................................................................... 102
4.8 Acknowledgements........................................................................................ 102
CHAPTER 5 THE TRITERPENOID CDDO ENHANCES DOXORUBICIN-
MEDIATED CYTOTOXICITY IN T-ALL CELLS................................................. 103
5.1 Abstract ......................................................................................................... 103
5.2 Introduction ................................................................................................... 104
xix
5.3 Methods..........................................................................................................108
5.3.1 Patients ....................................................................................................108
5.3.2 T-ALL cell lines ......................................................................................108
5.3.3 Cell culture ..............................................................................................108
5.3.4 Cell viability assay...................................................................................109
5.3.5 Measurement of apoptosis........................................................................109
5.3.6 Quantitative real-time reverse transcriptase PCR......................................110
5.4 Results............................................................................................................110
5.4.1 CFLAR expression is higher in specimens derived from patients at relapse
compared to the paired initial diagnostic specimen ...........................................110
5.4.2 Increased expression of CFLAR was associated with increased DOX
resistance..........................................................................................................112
5.4.3 CDDO reveals single agent cytotoxicity against T-ALL cell lines at
sub-micromolar concentrations .........................................................................112
5.4.4 Mechanism of cytotoxicity.......................................................................113
5.4.5 Effect of CDDO on CFLAR mRNA .........................................................115
5.4.6 CDDO enhances DOX-induced cytotoxicity ............................................116
5.5 Discussion ......................................................................................................117
5.7 Author contributions.......................................................................................122
5.8 Acknowledgements ........................................................................................122
CHAPTER 6. DISCUSSION ..................................................................................123
6.1 Challenges for childhood T-ALL ....................................................................123
6.2 Improving patient stratification - the genomic era ...........................................125
6.2.1 Microarray gene expression technology to identify prognostic markers....127
6.3 Towards improved outcome – identification of novel targets ..........................131
6.4 Future directions.............................................................................................136
xx
6.5 Summary........................................................................................................ 136
REFERENCES........................................................................................................ 138
1
CHAPTER 1
INTRODUCTION
1.1 Childhood T-cell acute lymphoblastic leukaemia
Acute lymphoblastic leukaemia (ALL) is the most common childhood malignancy and
accounts for one-quarter of all childhood cancers (Parkin et al, 2003). It is subdivided
into three groups, according to the surface immunophenotype of the cells,
pre-B ALL, mature B-cell ALL and T-cell ALL (T-ALL), the latter accounting for 10
to 15% of cases of ALL (Schrappe et al, 2000a, Gaynon et al, 2000).
Over the past four decades, the use of essentially empirically based therapy has resulted
in dramatic improvements in long-term survival, transforming a universally fatal
disease (Simone, 2003) into one where the rate of cure exceeds 80%
(Pui & Evans, 2006a). Historically, children with T-ALL were ascribed with a
significantly worse prognosis compared to children with pre-B ALL
(Uckun et al, 1996a; 1998). With current risk-adapted therapies, survival is now
comparable to pre-B ALL patients presenting with high risk features, with cure rates of
approximately 75% (Schrappe et al, 2000b; Asselin et al, 2001; Goldberg et al, 2003;
Pui et al, 2004a; Seibel et al, 2008). The reasons for these improved rates of survival
are due to the evolution of intensified risk-adapted therapeutic strategies developed
through randomised clinical trials, assisted by major improvements in supportive care
to reduce associated morbidity and mortality. It is worth noting that this progress has
not come about through the use of novel chemotherapeutic agents, but rather through
the improved use of existing ones, many of which were developed half a century ago.
However, over the past decade, survival has remained almost static (Summarised in
Table 1.1), with most groups still reporting 5 year event free survival (EFS) rates of
around 75% (Schrappe et al, 2000b; Asselin et al, 2001; Goldberg et al, 2003;
2
Pui et al, 2004a; Seibel et al, 2008). Moreover, for patients who relapse the outcome
remains dismal, despite the use of very intensive salvage regimes, including allogeneic
bone marrow transplantation (BMT) (Goldberg et al, 2003; Einsiedel et al, 2005). In
order to further improve cure rates several critical issues require resolution. Firstly,
therapeutic advances are urgently needed for children who relapse. Secondly, there is a
need for early identification of patients with the highest risk of relapse and once
identified, novel therapeutic strategies instituted to prevent relapse. Finally, there is a
need to identify the number of patients currently potentially overtreated and thus
unnecessarily subjected to severe acute toxicities and long term sequelae without
benefit, as this group of patients would benefit from reductions in therapy.
In the past decade substantial progress has been made in the understanding of
T-lymphoblast biology. The incorporation of biologic advances is expected to translate
into refinements in disease classification leading to better risk-stratification and
therapies. This chapter reviews the current management and biology of childhood
T-ALL.
3
Table 1.1 Results of selected clinical trials in T-ALL by comparison with pre-B ALL conducted over the last 20 years. Period study conducted over in parenthesis.
T-ALL pre-B ALL
Study # of Pts
5 yr EFS (unless otherwise stated)
# of Pts
5 yr EFS (unless otherwise stated)
Refs
AIEOP AIEOP-ALL-91 (1991-95)
144 40.4±4.1%
1050 74.9±1.4% 1,2
AIEOP-ALL-95 (1995-1999 )
34 66.6±8.2% (4yr EFS) result for PPR patients only
N/A N/A 3
BFM BFM-86 (1986 – 90)
127 71.3±4%
846 71.8± 1.6% 4,5
BFM-90 (1990-95)
284 61.1 ±2.9%
1828 80.4± 0.9% 5,6
CCG All protocols btw 1983-88
319 60±1%
1280 68±2% 7
All protocols btw 1989-95
431 73±2%
2883 75±1% 7
CCG-1961 (1996-2002)
235 72.3±6.2% vs 82.9±5.4% (Std vs stronger intensity)
880 70.4±3.4% vs 80.4±2.9% (Std vs stronger intensity)
8,9
DFCI All protocols btw 1981-1995
125 75±4%
1130 79±1% 10
DFCI 95-01 (1996-2000)
52 85±5% 434 81±2% 11
MRCUK ALLX (1985-1990)
139 48±4.2% 1349 64±1.3% 12
ALLXI (1990-1997)
207 51±3.5% 1730 65±1.2% 12
POG POG 8691 & 8704 (1986-1992)
439 51±2.4%
3828 70.9±0.8% 13,14
POG 9941 (1996-2000)
441
72.2±6.7% without HD-MTX vs 86±5.6% with HD-MTX
(3 yr EFS)
N/A
N/A
15
SJCRH Study XI (1984-88)
62 50.5±6.3%
296 76.4±2.5% 16
Study XII (1988-91)
29 51.7±9%
159 70.4±3.6% 16
Study XIIIA (1991-94)
23 60.9±10.2%
142 79.5±3.3% 16
Study XIIIB (1994-98)
43 71.9±6.8% 202 82.6±2.8% 17
Abbreviations: EFS, event-free survival; Std, standard; PPR, poor prednisone response; Pts, patients; vs, versus; AIEOP, Associazione Italiana Ematologia Oncologia Pediatrica; BFM, Berlin-Frankfurt-Münster; CCG; Children’s Cancer Group; DFCI, Dana-Farber Cancer Institute; MRCUK, Medical Research Council United Kingdom; POG, Paediatric Oncology Group; SJCRH, St Jude Children’s Research Hospital; N/A, not available; HD-MTX, high-dose methotrexate Refs: 1 Conter et al, 1998; 2 Conter et al, 2000; 3 Arico et al, 2002; 4 Reiter et al, 1994; 5 Schrappe et al, 2000a; 6 Schrappe et al, 2000b; 7 Gaynon et al, 2000; 8 Seibel et al, 2004; 9 Seibel et al, 2008; 10 Goldberg et al, 2003; 11 Moghrabi et al, 2007; 12 Eden et al, 2000; 13 Amylon et al, 1999; 14 Moloney et al, 2000; 15 Asselin et al, 2001; 16 Pui et al, 2000; 17 Pui et al, 2004a.
4
1.2 Clinical features
T-ALL is characterised by distinct clinical presenting features compared to children
with pre-B ALL. This includes generally older age (> 9 years), male gender
(3:1 male: female ratio), higher white blood cell (WBC) counts (> 50 x 109/L) and
higher rates of overt central nervous system (CNS) leukaemia (CNS-3; defined as > 5
WBC/μL CSF with blasts, in the absence of a traumatic lumbar puncture) at
presentation. Indeed, overt CNS leukaemia comprises approximately 11% of patients
presenting with T-ALL (Maloney et al, 2000; Burger et al, 2003), and as high as 20%
in one series (Goldberg et al, 2003), compared with approximately 2% for patients
presenting with pre-B ALL (Maloney et al, 2000; Burger et al, 2003;
Schultz et al, 2007). As a consequence, children with T-ALL are overrepresented in
National Cancer Institute (NCI) high risk groups defined by factors of age and WBC
count; only 25% of patients with T-ALL are classified as NCI standard risk compared
with 60% for pre-B ALL (Smith et al, 1996). Additionally, T-ALL frequently presents
with “bulky” or “lymphomatous” disease, which is defined as the presence of a
mediastinal mass, marked hepatosplenomegaly and/or lymphadenopathy
(Steinherz et al, 1998). Historically, the presence of many of these features at
presentation, for example lymphomatous disease and CNS involvement, was predictive
of an inferior outcome (Steinherz et al, 1998). However, with the use of contemporary
treatment strategies these features are no longer associated with patient outcome
(Attarbaschi et al, 2002; Goldberg et al, 2003).
Risk-adjusted therapy has become the hallmark of treatment for paediatric ALL, with
the goal of minimal treatment to attain a cure. This strategy relies on the presence of
reliable prognostic markers to guide therapy. Whilst pre-B ALL is characterised by
defined clinical and biological prognostic markers, in contrast, prognostic markers in
paediatric T-ALL are generally not well defined (Pullen et al, 1999;
5
Goldberg et al, 2003; Pui et al, 2004a). A notable example is the classification of
patients according to age and WBC (Smith et al, 1996). This classification has
repeatedly been shown to be highly effective for the stratification of pre-B ALL
(Vilmer et al, 2000; Eden et al, 2000) but has proven inadequate for T-ALL patients
(Pullen et al, 1999; Vilmer et al, 2000; Goldberg et al, 2003; Pui et al, 2004a).
1.3 Prognostic factors
1.3.1 Cytogenetics
For pre-B ALL, the presence of certain chromosomal abnormalities, most notably MLL
gene rearrangements in infants or the Philadelphia chromosome, are strongly associated
with poor disease-risk and are used to select treatment intensity. In contrast, there is a
lack of recurrent chromosomal abnormalities that predict prognosis in T-ALL patients
(Heerema et al, 1998), except perhaps for the small subgroups harbouring 10q24
abnormalities (Schneider et al, 2000) or paradoxically MLL gene rearrangements
(Rubnitz et al, 1999). However, several T-ALL specific transcription factors appear to
be useful for patient prognostication (see Biology section), but these have yet to be
prospectively validated and none are in current use for patient stratification.
1.3.2 Immunophenotype and T-lymphoblast maturational stage
Conflicting reports exist regarding the prognostic relevance of cluster differentiation
(CD) antigen expression. Perhaps the most consistent prognostic significance has been
associated with the expression of the CD2 antigen (Steinherz et al, 1986;
Gaynon et al, 1988a), where high CD2 expression appears to confer a favourable
prognosis (Uckun et al, 1996). Consistent with these reports, using gene expression
arrays to identify novel prognostic markers in adult T-ALL, Italian investigators
6
identified 3 genes predictive of outcome, one of which was CD2 (Chiaretti et al, 2004).
When combined with an abnormal karyotype, CD3 antigen expression was reported to
be associated with an inferior outcome in one series (Pui et al, 1990), but other studies
did not observe this (Shuster et al, 1990; Uckun et al, 1997). Similarly, the CD10
antigen (CALLA) has been associated with a favourable prognosis in some studies
(Dowell et al, 1987; Shuster et al, 1990), but not in a more recent Paediatric Oncology
Group (POG) analysis (Pullen et al, 1999).
Certain stages of T-lymphoblast maturational arrest have also been associated with
patient outcome. Patients whose T-lymphoblasts exhibit characteristic features of the
immature stage of thymocyte development (CD7+, CD2-, CD5-), have been shown to
have an adverse outcome (Uckun et al, 1997). On the other hand, T-lymphoblasts
displaying the cortical or intermediate immunophenotype (CD1a+) may have a superior
prognosis when treated with high risk based therapy (Ludwig et al, 1990;
Pullen et al, 1999), possibly due to an increased sensitivity to dexamethasone and
doxorubicin (Niehues et al, 1999; Wuchter et al, 2002).
1.3.3 In vivo response to therapy
Over the last two decades the in vivo response to chemotherapy has emerged as the
most powerful prognostic marker. The in vivo response to chemotherapy can be
assessed by morphological bone marrow (BM) assessment of blast clearance at day 7
or/and day 14 of induction (Children’s Oncology Group [COG] studies) or the
reduction of peripheral blood blasts after a 7 day course of prednisone and one
intrathecal (IT) methotrexate injection (Berlin-Frankfurt-Münster [BFM] studies). Both
methods predict patient outcome (Gaynon et al, 1990; Steinherz et al, 1996; reviewed
in Gaynon et al, 1997; Reiter et al, 2000; Schrappe et al, 2000a; 2000b) and have been
successfully used for early patient stratification (BFM-86, -90 and former Children’s
7
Cancer Group [CCG] studies). Interestingly, a higher incidence of poor prednisone
response (PPR) has been reported in children with T-ALL compared to those with
pre-B ALL, implying that T-lymphoblasts are relatively steroid resistant
(Schrappe et al, 2000b) which may be secondary to reduced glucocorticoid receptor
levels (Bell et al, 1983; Kato et al, 1993; Geley et al, 1996).
More recently, very sensitive molecular techniques (flow cytometry and quantitative
real time PCR [qPCR]) have been used to measure residual leukaemic cells well below
the level detectable by morphologic assessment. This is termed minimal residual
disease (MRD). These techniques allow the detection of as few as 1 leukaemic cell in
100,000 (10-5) normal cells, compared to less than 5 in 102 (< 5% blasts) by
morphologic assessment, and have been shown to be highly predictive of treatment
failure (Neale et al, 1991; Dibenedetto et al, 1997; Willemse et al, 2002;
Marshall et al, 2003) and are stronger prognostic markers than the prednisone response
(Van Dongen et al, 1998). Similar to the observation made regarding a higher incidence
of PPR among T-ALL patients, a significantly higher frequency of MRD-positive
patients, at most time points but especially at earlier time points has also been reported.
Moreover, the prognostic value of early MRD time points was also higher in T-ALL
patients (Willemse et al, 2002). MRD assessment has been incorporated into current
ALL studies to prospectively assess the prognostic impact of treatment modification
based on these measurements.
8
1.4 Treatment of childhood T-ALL
1.4.1 Historical perspective
Treatment strategies for T-ALL have closely mirrored those of its more common
counterpart, pre-B ALL. Historically, however, children with T-ALL have had
significantly worse outcomes than children with pre-B ALL, independent of the
unfavourable presenting features associated with T-ALL (Uckun et al, 1998;
Pui et al, 2004b). With improved treatment strategies the adverse prognostic
significance associated with the T-cell immunophenotype has significantly decreased
with EFS for T-ALL patients now approaching that for pre-B ALL (Uckun et al, 1996a;
1998; Seibel et al, 2008), summarised in Table 1.1. However, it is noteworthy that
childhood T-ALL displays unique features in response to treatment, including higher
rates of induction failure, shorter time to relapse and a 2 to 3-fold higher rate of CNS
relapse (Goldberg et al, 2003; Seibel et al, 2008). These characteristics further
highlight the distinct underlying biology between T-ALL and pre-B ALL and suggest
that T-lymphoblasts are intrinsically more resistant to chemotherapy
(Goldberg et al, 2003). This premise is supported by the observation that
T-lymphoblasts exhibit greater in vitro resistance to several anti-leukaemic drugs
including, glucocorticoids, vincristine and asparaginase (Pieters et al, 1993; 1998).
1.4.2 Treatment phases
Different therapeutic approaches have been utilised by the various groups that treat
childhood T-ALL; some groups have treated children with T-ALL on separate trials to
patients with pre-B ALL (former POG studies), whilst others have utilised the same
protocols, albeit on more intensive higher risk treatment arms (former CCG,
Dana-Farber Cancer Institute (DFCI), BFM groups). Most groups utilise the same
9
treatment schema for T-ALL as the one devised for treating pre-B ALL. As predicted
by the Goldie-Coldman model (Goldie and Coldman, 1979), effective strategies to treat
childhood T-ALL have included exposure to multiple non-cross resistant cytotoxic
agents, delivered early to prevent the emergence of tumour resistance. To provide
control of CNS leukaemia, systemic therapy must be used in combination with
prophylactic CNS directed therapy, including cranial radiotherapy (CRT), as discussed
in detail below and IT therapy. To date, the most successful protocols have adopted the
principle of early treatment intensification (see below). These approaches have resulted
in a progressive increase in survival for childhood T-ALL, summarised in Table 1.1.
Generally, ALL therapy is divided into several phases including induction,
consolidation, re-induction (also known as delayed intensification) and continuation
(also known as maintenance) phases.
1.4.2.1 Induction
Three to four anti-leukaemic agents are typically used in induction. Patients classified
as standard risk receive three drugs consisting of a corticosteroid (usually prednisone or
dexamethasone), vincristine and asparaginase. In addition to these three anti-leukaemic
drugs, high risk patients also receive an anthracycline (daunorubicin or doxorubicin).
This combination generally results in remission in 88 to 95.5% of T-ALL patients. By
comparison, 98 to 98.5% of pre-B ALL patients generally achieve remission by the end
of induction (Goldberg et al, 2003; Pui et al, 2004a).
1.4.2.2 Consolidation
Following remission induction therapy, patients are treated with consolidation therapy,
which often includes asparaginase, methotrexate (often given in high dose)
10
(Schrappe et al, 2000b; Pui et al, 2004a) and a thioguanine (either mercaptopurine or
more typically 6-thioguanine). Some groups (COG) also include an alkylator
(cyclophosphamide) and cytarabine (also known as Ara-c) during this phase. CNS
directed therapy, which consists of CRT (see CNS directed therapy section for full
details) and/or IT chemotherapy is delivered during consolidation. IT chemotherapy
consists of either methotrexate alone or methotrexate combined with cytarabine and
hydrocortisone, which is termed triple intrathecal (TIT) therapy.
1.4.2.3 Re-induction and interim maintenance
The introduction of the re-induction phase in the 1970s, developed by Rhiem and
Henze (Rhiem et al, 1980; Henze et al, 1981; 1982) (BFM-76/79 study) from the BFM
consortium, had a profound impact on the cure rate for childhood T-ALL, improving
survival from approximately 40% to 70% (Rhiem et al, 1980; Henze et al, 1981; 1982).
This therapeutic approach, which is also known as re-intensification or delayed
intensification, is essentially a repetition of induction and consolidation therapy,
delivered again early in remission. Many other cooperative groups also adopted this
strategy and confirmed its effectiveness (Gaynon et al, 1988a; 1988b; 1993;
Pui et al, 2004a).
Re-induction is generally preceded by a course of non-myelosupressive therapy, termed
interim maintenance, delivered in an effort to allow marrow recovery, whilst continuing
exposure to multiple anti-leukaemic agents. This phase, which initially consisted of low
dose methotrexate and mercaptopurine, has been progressively intensified, and now
includes multiple courses of asparaginase and vincristine administered in conjunction
with high-dose methotrexate (European Groups) (Reiter et al, 1994) or Capizzi style
methotrexate (COG) (Nachman et al, 2005). The inclusion of a re-intensification phase
11
(ALL-BFM-86 study) (Reiter et al, 1994) was found to be particularly beneficial for
the subgroup of T-ALL patients who had a prednisone good response (PGR). These
patients achieved a 6 year EFS of 84%±4%. However, despite additional treatment
intensification in consolidation, this strategy failed to improve outcome for T-ALL
patients with a PPR, who only had a 6 year EFS of 45%±9% (Reiter et al, 1994).
It is also clear that the delivery style of therapy significantly impacts on patient
outcome (Reiter et al, 1994). Intensive therapy delivered in alternating blocks, as done
in the ALL-BFM-90 study, (Schrappe et al, 2000b), achieves inferior results to therapy
delivered in a more continuous fashion (Reiter et al, 1994; Arico et al, 2002). The
reason for this may be due to longer pauses without therapy using block style
approaches.
1.4.2.4 CNS directed therapy
Prior to the introduction of CNS directed therapy up to two thirds of children with ALL
succumbed to a relapse involving the CNS (Evans et al, 1970). The recognition that the
CNS provides a sanctuary site for leukaemic blasts prompted investigators to introduce
CNS directed therapy to treat occult CNS leukaemia. CNS prophylaxis was initially
achieved using 24Gy of radiotherapy delivered to the whole neuroaxis, in combination
with IT methotrexate (Aur et al, 1971; Hustu et al, 1973). These studies and others that
followed (Rivera et al, 1993) demonstrated the high efficacy of this approach for
preventing CNS leukaemia relapse. However, due to the sensitivity of the developing
CNS to radiation, children treated with CNS radiotherapy sustained damage to normal
brain, resulting in significant deleterious sequelae to neurocognitive, neuropsychologic
and neuroendocrine function (Mulhern et al, 1991; Jankovic et al, 1994; reviewed in
Robison and Bhatia, 2003) as well as the development of secondary CNS tumours
(Packer et al, 1987; Walter et al, 1998; Pui et al, 2003). These sequelae were
12
particularly devastating in younger children, especially those aged less than 5 years at
the time of treatment, prompting investigators to adopt alternative strategies to CRT.
The complete omission of CRT, without detrimental effects on outcome, has been
successfully achieved for patients with lower and intermediate risk pre-B ALL, by
using intensified IT therapy in combination with intensive systemic therapy
(Pullen et al, 1993; Conter et al, 1995; Pui et al, 2004a). However, due to the increased
risk of CNS relapse ascribed to T-ALL patients (Steinhertz et al, 1991;
Smith et al, 1996), especially those presenting with a WBC count above 100 x 109/L
(Conter et al, 1997), many investigators have been reluctant to omit CRT for children
with T-ALL. A prospective randomised study on a backbone of contemporary intensive
systemic and IT chemotherapy, would definitively answer the question of the role of
CRT in T-ALL. However, it has been estimated that to perform such a study would
require 1440 patients (Pui et al, 2001), thus requiring international participation of
cooperative groups. The large meta-analysis, conducted by Clarke et al (2003),
revealed that long-term IT therapy gave similar outcomes to CNS-directed
radiotherapy. Additionally, no significant difference in outcome was found according
to the administration of different CRT doses. However, distinct conclusions could not
be made with regard to T-ALL.
Studies testing the feasibility of omitting CRT have yielded mixed results
(Laver et al, 2000; Vilmer et al, 2000). The former POG reported a comparison of CRT
versus no CRT, from data derived from a series of six POG trials conducted between
1987 and 1995, which investigated a total of 222 children with T-ALL
(Laver et al, 2000). Though there were no statistically significant differences in EFS
rates between the two groups (65% with CRT compared with 63% without CRT),
patients who did not receive CRT had a significantly higher CNS relapse rate compared
to patients who received CRT (18% compared with 7%) (Laver et al, 2000). On the
13
other hand, investigators from the European Organisation for Research and Treatment
of Cancer (EORTC) (Vilmer et al, 2000) revealed no difference in the incidence of
CNS relapse or overall patient outcome, for intermediate and high risk patients,
following omission of CRT on a BFM backbone, which included high-dose
methotrexate (5g/m2) and IT methotrexate for CNS prophylaxis (EORTC 58832 study).
It is noteworthy however, that immunophenotypic data were not available for this
study. In their subsequent study (EORTC 58881), the addition of high-dose Ara-c to the
same regimen did not improve outcome (Millot et al, 2001). In this study, the
cumulative CNS relapse rate at 8 years for T-ALL was 13.3%±2.3% compared with
7.6%±0.8% for pre-B ALL, revealing that even with intensified systemic therapy,
which included high-dose methotrexate, patients with T-ALL continue to have a higher
risk of CNS relapse compared to patients with pre-B ALL. The former CCG recently
reported the results of the CCG-1961 study (Seibel et al, 2008). CRT was omitted for
all NCI high risk patients, including all T-ALL patients, with a rapid early response
(RER), which is defined as <25% blasts in BM at day 7, to induction therapy. Although
favourable results were reported for patients with T-ALL, it is noteworthy that two
thirds of all relapses in this group were isolated in the CNS (Seibel et al, 2004). This
finding prompted the reintroduction of CRT in the current frontline COG T-ALL
protocol (AALL0434) for NCI high risk T-ALL patients with a RER
(Seibel et al, 2004).
Another strategy employed by other cooperative groups to reduce the sequelae of CRT
has been to reduce the doses of CRT delivered. The dose of CRT has been successfully
reduced from 24Gy to 18Gy and more recently 12Gy, without an increase in CNS
events (Schrappe et al, 2000b). This study revealed an extremely low isolated CNS
relapse of 0.8% and 1.6% for medium risk (n=1299 of which 13.8% were T-ALL
patients) and high risk (n=243 of which 44.8% were T-ALL patients) patients,
14
respectively. Separate results were not provided for T-ALL patients. Based on these
results, the current BFM ALL study (ALL-BFM-2000) utilises 12Gy CRT for all
T-ALL patients.
A subgroup of T-ALL patients at reduced risk of CNS relapse includes patients
presenting with a WBC count less than 100 x109/L (Conter et al, 1997;
Schrappe et al, 2000b; Arico et al, 1997). Additionally, T-ALL patients without
lymphomatous features at presentation and a RER appear to be at reduced risk of CNS
relapse and do not require prophylactic CRT (Nachman et al, 1997).
In the St Jude Total Therapy XIIIB trial (Pui et al, 2004a), CRT was only administered
to patients with T-ALL if they had a WBC count greater than 100 x109/L (18Gy) or to
patients who had a CNS-3 status (24Gy). A total of 247 patients were enrolled
(T-ALL n=43), of which 30 received CRT. The number of T-ALL patients who
received CRT was not included. The remainder of patients were treated with early
intensification of TIT therapy. These investigators reported an encouragingly low rate
of isolated CNS relapse (1.7%) and overall CNS relapse (3%) at 5 years. For children
with T-ALL the overall CNS relapse rate was 7% (3 of 43 patients: 1 isolated and 2
combined CNS and BM) (Pui et al, 2004a), however, a number of these children would
have received CRT. St Jude investigators attributed the success of the strategy to the
early use of intensified TIT therapy combined with effective systemic therapy
(dexamethasone instead of prednisone and earlier use of re-induction therapy)
(Pui et al, 2004a). Based on these results St Jude investigators have entirely removed
CRT from their recently closed frontline ALL study (St Jude Total Therapy XV),
which included children with T-ALL, reserving CRT as salvage therapy for patients
with CNS relapse (Pui, 2006c).
15
The use of TIT also provided superior CNS prophylaxis in a study conducted by the
former CCG (CCG-1952) (Matloub et al, 2006) between 1996 and 2000 for standard
risk ALL patients. In this study patients were randomised between intensified TIT and
IT methotrexate alone. Although the study revealed a significant decrease in CNS
relapse rate with the use of intensified TIT, an increase in haematologic and testicular
relapses was observed, especially among T-ALL patients, resulting in a decrease in
overall survival (OS) in this treatment group (Matloub et al, 2006), revealing that for
this approach to be effective, adequate systemic therapy is required.
1.4.2.5 Continuation
The final phase of treatment is termed continuation and consists of anti-metabolite
agents with pulses of vincristine and corticosteroids for two to three years depending on
the gender of the patient, with boys generally receiving longer therapy than girls.
Attempts to reduce the length of this phase of therapy by 6 months have yielded
significantly reduced EFS rates (Tsuchida et al, 2000).
1.4.3 Increasing dose intensity
Investigators from the former CCG conducted a study, CCG-1882,
(Nachman et al, 1997), to determine if an intensified BFM regimen, termed
augmented-BFM, could abolish the adverse prognostic effect of a slow early response
(SER), which is defined as >25% blast on day 7 BM assessment. The regimen included
two re-induction phases and intensive use of non-myelosuppressive agents, including
Capizzi style methotrexate, in interim maintenance. Though the study only included a
limited number of children with T-ALL (n=22), all without lymphomatous features,
this group of patients responded especially favourably to this regimen, with an
16
excellent 5 year EFS of 90.7%±7.1% (Nachman et al, 1997). By comparison,
pre-B ALL patients had a 5 year EFS of just over 50% (P = 0.008)
(Nachman et al, 1997). The authors ascribed the success of this strategy to the delivery
of increased dose intensity of non-myelosuppressive agents (vincristine, asparaginase
and corticosteroids) to patients (Nachman et al, 1997). Notably, significantly more of
these drugs were administered compared to the ALL-BFM-86 and 90 studies
(Schrappe et al, 2000b). This treatment strategy was however, associated with a high
rate of avascular necrosis of bone (Mattano et al, 2000). Paradoxically, T-ALL patients
with a RER (n=17) who were treated on the same protocol (CCG-1882) but using
standard CCG-modified BFM therapy, only achieved an EFS of 63.7%±11.9% at
5 years (Nachman et al, 1997). This suggests that T-ALL patients, without
lymphomatous features and a RER to induction chemotherapy, are likely to benefit
from additional treatment intensification comparable to the approach used in patients
with a SER.
This hypothesis was tested in the subsequent CCG study for high risk ALL patients,
CCG-1961 (Seibel et al, 2008), conducted between 1996 and 2000. This study
randomised 263 patients with T-ALL, NCI high risk features and a RER, to receive
augmented therapy with or without two re-induction phases in a four arm strata
(Seibel et al, 2008). Patients treated on the augmented intensity regimen had the best
outcome with 5 year EFS and OS of 76.9% and 82.5% respectively (Seibel et al, 2008).
There was no difference in outcome according to number of re-induction phases
delivered (one delayed intensification phase compared with two delayed intensification
phases). Importantly, this result was achieved without the use of prophylactic CRT.
The use of discontinuous dexamethasone in this protocol during delayed intensification
also significantly reduced the rate of avascular necrosis of bone (Mattano et al, 2003).
17
1.4.4 Chemotherapeutic agents
In addition to the strategy employed, certain anti-leukaemic agents appear to be
particularly efficacious at treating T-ALL. These include anthracyclines, asparaginase,
methotrexate, dexamethasone and 6-thioguanine.
1.4.4.1 Anthracyclines and asparaginase
The intensive use of anthracyclines and asparaginase has formed the backbone of DFCI
protocols, which have consistently produced some of the best outcomes for childhood
T-ALL. Indeed, their most recently published results using the DFCI 95-01 protocol,
which ran from 1996 to 2000, reported the best overall outcome for unselected T-ALL
patients, with survival rates of 85%±5% at 5 years (Moghrabi et al, 2007). At DFCI,
children with T-ALL are automatically categorised as high risk and treated on high risk
protocols (Goldberg et al, 2003; Moghrabi et al, 2007), which include the use of
prophylactic CRT (18Gy). POG studies also showed improved outcome with the
intensive use of asparaginase (Amylon et al 1999).
1.4.4.2 Methotrexate
The folic acid antagonist methotrexate plays an integral role in the treatment of T-ALL.
Methotrexate competitively and reversibly inhibits dihydrofolate reductase (DHFR),
thus interfering with DNA synthesis. The optimal dose of methotrexate in the treatment
of T-ALL has yet to be determined (Evans et al, 1999), despite a multitude of studies
attempting to specifically address this issue. Inter-study comparison is problematic for
many reasons, due to the use of a wide range of dosages (20mg/m2 to 35g/m2) and
infusion lengths of methotrexate used to treat children with ALL.
18
Several lines of evidence support the potential efficacy of high-dose methotrexate in
preference to lower doses of methotrexate, for the treatment of childhood T-ALL.
Higher levels of the active long chain metabolites, methotrexate polyglutamates
(MTXPG), have been shown to result in increased in vivo cytotoxicity
(Masson et al, 1996). Due to lower activity of the enzyme folylpolyglutamate
synthetase (FPGS) (Barredo et al, 1994; Galpin et al, 1997), children with T-ALL
accumulate lower concentrations of MTXPG (Masson et al, 1996; Kager et al, 2005).
This has recently been shown to be due to lower levels of the FPGS gene in patients
with T-ALL (Kager et al, 2005). Additionally, T-lymphoblasts have higher levels
of DHFR compared with pre-B lymphoblasts (Matherly et al, 1997), which is also
linked with lower MTXPG levels. Taken together, the above data provide a rationale
for the use of high-dose methotrexate to overcome methotrexate resistance mediated by
the above factors (Barredo et al, 1994). Moreover, there is a high degree of variability
with regards to the pharmacokinetics (e.g. systemic clearance) of high-dose
methotrexate, which has also been shown to affect EFS of paediatric patients with ALL
(Borsi and Moe, 1987a; 1987b). Children with more rapid systemic clearance of
high-dose methotrexate, leading to lower systemic exposure to methotrexate, had
higher rates of relapse (Borsi and Moe, 1987a) and inferior outcome
(Seidel et al, 1997).
Several cooperative groups have increased the doses of methotrexate administered to
patients with T-ALL, in an attempt to improve outcome. However, this approach has
not been universally successful, possibly because MTXPG accumulation is saturable
(Borsi et al, 1990). The Childhood ALL Collaborative Group conducted a large
meta-analysis, which included over 9,000 children enrolled on 43 randomised trials that
began before or during 1993 (Clarke et al, 2003). Studies using high-dose methotrexate
(0.5g/m2 to 8g/m2) in combination with an assortment of leucovorin rescue schedules
19
revealed a significant reduction in non-CNS relapse rates by 17%, improved EFS rates
(68.1% versus 61.9% P = 0.003) and a trend toward better OS at 10 years
(80.1% versus 76.8% P = 0.09) (Clarke et al, 2003). CNS relapse rates were decreased
by 19%, however, this difference did not reach statistical significance (P = 0.08)
(Clarke et al, 2003). Due to a lack of immunophenotypic data, distinct conclusions
could not be made with regard to T-ALL. Additionally, there was insufficient data to
determine whether higher doses of methotrexate were superior to lower doses
(Clarke et al, 2003).
The POG-9404 study was the first randomised trial addressing the efficacy of
high-dose methotrexate in children with T-ALL (Asselin et al, 2001). The study was
designed to determine if the addition of high-dose methotrexate (5g/m2) improved
outcome of the DFCI 87-01 protocol (Silverman et al, 2000). The study enrolled 441
children and preliminary analysis revealed a 3 year EFS of 86±5.6% for patients
receiving high-dose methotrexate compared with 72.2%±6.7% for patients who did not
receive high-dose methotrexate. Significant decreases in induction failures and reduced
isolated CNS relapses accounted for the improved EFS (Asselin et al, 2001). As a
result of this analysis the study was closed to accrual early. Final analyses of this study
are awaiting publication.
A retrospective analysis of 26 children with T-ALL less than 5 years old, demonstrated
equivalent survival, importantly with reduced late effects, for patients who received
high-dose methotrexate versus those who were irradiated (Nathan et al, 2004). The
authors concluded that CNS CRT can be substituted by high-dose methotrexate
(5g/m2 x 4 over 24 hours) in this population (Nathan et al, 2004).
A series of three pilot trials were carried out by the CCG and NCI between 1977 and
1991 to ascertain the efficacy of very high-dose methotrexate (33.6g/m2), as an
20
alternative CNS prophylactic strategy to CRT. The first study CCG-191P, which ran
from 1977 to 1983, randomised 181 patients, including 26 with T-ALL. High risk
patients, defined by WBC count > 50 x 109/L, treated on the methotrexate arm had a
very high rate of CNS relapse; 44%±9% in the methotrexate arm compared with
14%±5% in the CRT arm. The higher than expected rate of CNS relapses may at least
be partially explained by the early use of leucovorin rescue (Pui et al, 2006b), which
has been shown to counteract the anti-tumour effects of methotrexate
(Browman et al, 1990). Other proposed contributing factors for the very high rate of
CNS relapse observed include the use of insufficient CNS prophylaxis during the
induction phase and the use of prednisolone instead of dexamethasone
(Pui et al, 2006b). In the successor study CCG-134P conducted for high risk ALL
patients (total n = 128; T-ALL n = 47), intensified systemic and IT chemotherapy were
added to the very high-dose methotrexate regimen which resulted in a marked
reduction in CNS relapse rate (any CNS relapse rate 11%±3%). However, the 5 year
OS and EFS rates were unchanged from the preceding study, due to an increase in BM
relapses (5 year EFS 51% for both CCG-191P and CCG-134P P=0.99; 5 year OS 63%
on CCG-134P compared with 75% on CCG-191P P=0.1). The reason for the lack of
improvement in outcome may relate to the block style delivery of anti-leukaemic
agents during intensification, as discussed above (Schrappe et al, 2000b;
Pui et al, 2006b). Therefore, although well tolerated, administering very high-dose
methotrexate appears to provide no additional therapeutic benefit over the use of
high-dose methotrexate. This could possibly be due to the presence of a saturable
carrier system for methotrexate and its metabolite 7-OH methotrexate between serum
and CSF (Borsi et al, 1990).
21
1.4.4.3 Dexamethasone
Substituting dexamethasone for prednisolone has been shown to result in improved
EFS for both standard (Bostrom et al, 2003; Mitchell et al, 2005) and high risk ALL
patients (Mitchell et al, 2005). Although the main effect was via reductions in CNS
relapse (Mitchell et al, 2005), non-CNS relapse was also significantly reduced in the
patients treated with dexamethasone. This is consistent with the findings that
dexamethasone has superior CNS penetration (Balis et al, 1987) and a 16-fold higher
in vitro anti-leukaemic activity compared to prednisolone at 40mg/m2, as assessed in
3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) assays from 133
untreated ALL patients (Kaspers et al, 1996). However, it is noteworthy, that in a study
which compared dexamethasone (6mg/m2) to higher doses of prednisone (60mg/m2),
no benefit to dexamethasone was found (Igarashi et al, 2005), putting into the question
the equivalency of doses of dexamethasone compared to prednisone in previous
reports.
1.4.4.4 The anti-metabolite 6-thioguanine
The anti-metabolite 6-thioguanine also appears to significantly reduce the risk of
isolated CNS relapse compared to its thiopurine counterpart 6-mercaptopurine.
However, an increase in remission deaths, which were largely due to infective episodes
during continuation therapy, neutralised this advantage (Vora et al, 2006).
Interestingly, a study using MTT assay to investigate the in vitro sensitivity of T-ALL
blasts at relapse found that these blasts were highly sensitive to high doses of
thiopurines (Kaspers et al, 2005; Pui, 2005).
22
1.5 Future strategies
Attempts to successfully reduce therapy for childhood T-ALL have been hampered by
a lack of reliable prognostic markers. Protocols which have stratified T-ALL patients
according to NCI criteria and reduced therapy for T-ALL patients with NCI standard
risk features (WBC count < 50 x 109/L and age < 10 years) have demonstrated inferior
outcomes compared to pre-B ALL patients (Matlaoub et al, 2006). Strikingly, T-ALL
patients with NCI standard risk features treated according to the standard risk protocol,
CCG-1991, had similar outcomes to T-ALL patients presenting with high risk features
who were treated on the high risk protocol (CCG-1961); 5 year EFS of 73% for
standard risk compared with 72% for high risk (P = 0.77) (Matlaoub et al, 2006). By
contrast, T-ALL patients with standard risk features treated on a more intensive
protocol (POG-9404) fared much better; the estimated 5 year EFS for patients with
standard risk T-ALL on POG-9404 was 88%. These patients also had significantly
better outcomes compared to T-ALL patients with high risk features (5 year EFS of
90% for standard risk compared with 75% for high risk, P < 0.004)
(Matlaoub et al, 2006). Furthermore, the DFCI, which classifies all T-ALL patients as
high risk, regardless of age and presenting WBC count, has reported excellent and
similar 5 year EFS for T-ALL patients grouped according to NCI risk criteria
(83%±11% for standard risk compared with 85%±6% for high risk)
(Moghrabi et al, 2007). These results suggest that to achieve outcomes similar to pre-B
ALL patients, T-ALL patients presenting with NCI standard risk features require more
intensive therapy. These data further stress the importance of adequate patient
stratification.
For selected subgroups of T-ALL patients, therapeutic reductions have been possible.
Patients with a PGR were shown to have similar EFS to historical controls, following
23
therapy with reduced dose of 12Gy CRT and less anthracyclines (ALL-BFM 90)
(Schrappe et al, 2000b).
Over the past 3 decades chemotherapeutic agents have been extensively evaluated for
the treatment of T-ALL in various doses, combinations and schedules. Studies clearly
revealed that early treatment intensification provides the most effective strategy to treat
T-ALL. Additionally, certain cytotoxic agents appear to be especially efficacious in
the treatment of T-ALL, including doxorubicin (Goldberg et al, 2003), high-dose
asparaginase (Amylon et al, 1999; Goldberg et al, 2003), and systemic high-dose
methotrexate (Asselin et al, 2001). In addition, dexamethasone appears to provide
improved CNS leukaemia control (Bostrom et al, 2003; Mitchell et al, 2005). The use
of CRT for all patients with T-ALL is questionable in an era of intensified systemic and
IT therapy. It is clear that omitting CRT without adequate systemic chemotherapy
increases the rate of treatment failures, particularly in the neuroaxis (Laver et al, 2000).
Intensified TIT therapy, when used with effective systemic therapy provides a
successful CNS preventative strategy, negating the need for prophylactic CRT for
many patients with T-ALL (Pui et al, 2004a). The doses of CRT delivered have been
reduced without increases in CNS relapse rates, with 12Gy as effective as 18Gy
(Schrappe et al, 2000b). Patients with T-ALL presenting with a
WBC >100 x 109/L appear to be at an especially high risk of CNS relapse. This group
should continue to receive CRT (12Gy) until an alternative strategy of CNS
prophylaxis can be validated.
A significant number of children with T-ALL remain incurable with current therapies
and effective chemotherapy regimens remain elusive for the majority of T-ALL
patients who relapse. A plateau in efficacy has been reached using conventional
non-specific chemotherapy. Although modest improvements in outcome may be
achieved by further refining treatment schedules and introducing new chemotherapeutic
24
agents, further treatment intensification using non-specific chemotherapy is more likely
to result in additional toxicity without major advances in survival. To improve outcome
for patients with T-ALL, it is critical to better understand the tumour biology. Over the
past decade remarkable advances have been made in knowledge of T-ALL biology,
which has shifted the focus onto novel agents that target molecular changes critical for
T-lymphoblast proliferation and/or survival. These selective agents are predicted to be
less toxic to normal cells and it is anticipated that they will be more effective than
currently used chemotherapeutic agents. In the second part of this chapter the biology
of childhood T-ALL is reviewed with an emphasis on novel molecular classifiers.
1.6 Biology and new prognostic markers
The heterogeneity of T-ALL is underscored by the strikingly distinct outcomes
displayed by patients with similar clinical features at diagnosis, treated on the same
therapeutic protocols, suggesting that T-ALL is composed of a group of biologically
distinct diseases, which share the T-cell immunophenotype.
1.6.1 Origins of leukaemia – the cancer stem cell theory
This theory proposes that a minority population of self-sustaining cancer cells
(typically 0.1% of the tumour burden at diagnosis [Dick, 1996]) give rise to the bulk of
the tumour, which appears to have a more differentiated immunophenotype. These
cancer stem cells are capable of self-renewal and maintaining the mass of the tumour. It
is postulated that cancer stem cells retain other features of normal stem cells such as
low rates of cell division. Thus, current non-specific cytotoxic therapies may be
effective against the bulk of leukaemic blasts but may not eradicate the cancer stem
25
cells, ultimately culminating in disease recurrence (reviewed in Reya et al, 2001;
reviewed in Warner et al, 2004).
Although the existence of cancer stem cells was proposed many decades ago
(Bruce et al, 1963), the first evidence of the existence of a leukaemic stem cell (LSC)
came in 1994 (Lapidot et al, 1994). Lapidot et al (1994), isolated human acute myeloid
leukaemia (AML) cells using fluorescence activated cell sorting (FACS) and
demonstrated that only the CD34+ve CD38-ve fraction of cells were capable of initiating
leukaemia following transplantation into severe combined immune deficiency (SCID)
mice. More differentiated cells (CD34+ve CD38+ve) failed to give rise to AML. The
authors revealed that this population of cells demonstrated the property of self-renewal
by demonstrating serial-transplantation into nonobese diabetic (NOD)/SCID mice.
Thus, leukaemias are propagated by small fractions of LSCs. It appears that like the
well characterised hierarchy of normal haematopoiesis, leukaemia cells also exhibit a
hierarchy (Bonnet and Dick, 1997). The clinical implications of the existence of LSCs
are far reaching and are likely to result in a major change in the way we treat ALL.
Therapies specifically targeting LSCs may replace current non-specific therapies
(reviewed in Pardal et al, 2003; and in Huntly and Gilliland, 2005a; 2005b). A critical
factor will be to take advantage of differences between LSC and normal haematopoietic
stem cells. Proof of principle that LSCs may be selectively targeted has been
demonstrated (Yilmaz et al, 2006). It remains to be determined wether LSCs arise from
normal haematopoietic stem cells or a more differentiated progenitor, which has
acquired stem cell-like properties. It is important to note that whilst the cancer stem cell
theory is gaining wide acceptance especially for AML (Lapidot et al, 1994), brain
tumours (Singh et al, 2004) and breast cancer (Al-Hajj et al, 2003), some investigators
have challenged the paradigm (Kelly et al, 2007). Recent evidence suggests that the
existence of cancer stem cells may actually be disease specific (le Viseur et al, 2008).
26
This study revealed that self-renewal capacity was present in leukaemic lymphoid
progenitors of different stages of maturation and not restricted only to the most
immature cells (le Viseur et al, 2008). Thus, the cancer stem cell theory may not be
applicable to all cancers.
1.6.2 The putative T-ALL cell of origin
The characteristics of the T-ALL stem cell remain to be determined. It is hypothesised
that developing thymocytes arrested at various stages of thymocyte development are
the cells of origin of molecularly distinct subgroups of T-ALL
(reviewed in Rabbits 1991; reviewed in Armstrong and Look, 2005;
reviewed in Grabner et al, 2006; reviewed in Graux et al, 2006). To fully appreciate
events leading to the development of T-ALL, an understanding of normal
T-lymphocyte differentiation and maturation in the thymus is required.
1.6.2.1 Normal thymocyte development
Mature T-lymphocytes play a critical role in cell mediated immunity. Normal
T-lymphocyte production occurs in the thymus, via a process of hierarchical
development, starting with common lymphoid progenitors (CLP) which originate from
haematopoietic stem cells in the BM or liver. CLPs progress through multiple stages of
differentiation and proliferation, generating the relevant T cell receptors (TCR) on
mature thymocytes, which are needed to recognise foreign and self-antigens
(summarised in Figure 1.1). The earliest stage of T-cell development is represented by
the CD4/CD8 double negative (DN) (immature /pro-T/cytoplasmic) stage. This stage is
characterised by the presence of CD7 and CD34 and the absence of other
T-lymphocyte markers (CD1a-/CD3-/CD2-). DN thymocytes undergo so called beta
27
selection, involving structural rearrangement of the TCRβ chain. This process is
comparable to immunoglobulin gene rearrangement during B-lymphocyte
development. Successful rearrangement leads to up regulation of CD4 and CD8
expression on the cell surface, and they differentiate into double positive (DP)
thymocytes. As a result of a multitude of gene rearrangements, immature thymocytes
display an extensive range of TCRs (reviewed in von Boehmer et al, 2003).
Thymocytes failing β-selection undergo apoptosis. This early cortical
(intermediate/thymic) thymocyte stage is characterised by CD1a and CD10 expression.
During the late cortical stage of thymocyte differentiation thymocytes undergo
rearrangement of the TCRα chain locus, culminating in cell surface expression of
TCRα/β and CD3 (reviewed in von Boehmer, 2005). Subsequent developmental
decisions are dictated by interactions with peptides/major histocompatibility complex
(MHC) on stromal cells within the thymus microenvironment. The affinity/avidity of
this interaction determines the outcome of the selection process. Thymocytes with non-
functional TCR are unable to bind to MHC peptides and as a consequence do not
receive survival signals and die by neglect. On the other hand, the process of negative
selection removes autoreactive thymocytes, which bind with excess affinity to
peptide/MHC (reviewed in von Boehmer et al, 2003). The resultant mature single
positive (SP) thymocytes expressing either CD4 or CD8 on the cell surface are released
into the circulation and populate peripheral lymphoid tissue.
28
Figure 1.1 Normal T-lymphocyte development. As thymocytes differentiate and
mature in the thymus, they loose expression of CD34 whilst gaining CD4 and CD8
expression (adapted from De Keersmaecker et al, 2005; Graux et al, 2006)
1.6.2.2 T-cell development gone awry
Akin to other neoplasms, T-ALL develops as a result of a collection of mutations,
which occur in a cell, resulting in its transformation. In ALL, the initiating event is
often a chromosomal translocation (reviewed in Greaves, 2003). In most cases the
presence of a chromosomal translocation alone is insufficient to generate leukaemia;
additional mutations are required for malignant transformation.
Molecular analysis of T-lymphoblasts has revealed that T-ALL, like pre-B ALL is
characterised by a large number of cytogenetic abnormalities. Conventional
karyotyping detects non-random, recurring chromosomal translocations in
29
approximately 50 to 60% of T-ALL patients (Heerema et al, 1998;
Schneider et al, 2000). These translocations preferentially affect the TCR loci on
chromosomes 7q34 (TCRβ) and 14q11 (TCRα/δ) which juxtapose several
developmentally important transcription factors, including TAL1 (SCL), TAL2,
bHLHB1, LYL1, MYC (basic helix-loop-helic gene family) LMO1, LMO2, (LIM
domain only zinc finger encoding genes) HOX11 (TLX1), HOX11L2 (TLX3), HOXA
cluster (homeobox gene family), and MYB under the transcriptional control of potent
promoter and enhancer elements of the TCR genes, causing aberrant transcription factor
expression. Alternatively, for a smaller proportion of patients, chromosomal
translocations result in the creation of fusion proteins, including MLL-ENL,
CALM-AF10 and NUP214-ABL1. Many patients also harbour cryptic translocations,
detectable only by sensitive molecular techniques, such as fluorescent in situ
hybridisation (FISH) or quantitative real time reverse transcriptase polymerase chain
reaction (qRT-PCR). Moreover, it is evident that many of the above mentioned
transcription factors can also be activated in the absence of any detectable cytogenetic
abnormalities (Ferrando et al, 2002; Kees et al, 2003a), via yet to be elucidated
alternative trans-acting mechanisms. One possible mechanism proposed involves the
disruption of genes which normally control these transcription factors during
thymocyte development, resulting in their aberrant expression (Ferrando et al, 2004a).
This hypothesis was based on the observation that biallelic activation of TAL1, HOX11
and LMO2 expressing T-ALL samples was seen in 42%, 17% and 64% of cases
analysed respectively (Ferrando et al, 2004a; reviewed in O'Neil and Look, 2007).
The mechanisms leading to translocation formation, particularly where TCR genes are
involved, are thought to occur as a result of errors in the normal TCR gene
rearrangement process that occurs during thymocyte maturation, which places the
various transcription factor genes erroneously under the control of the potent promoter
30
and enhancer elements of the TCR genes (reviewed in Rabbits, 1993; reviewed
in Lichty et al, 1995; Salvati et al, 1999). The resultant aberrant expression of the
various transcription factors is thought to initiate T-cell leukaemogenesis and maintain
the leukaemic clones, by as yet undefined mechanisms. It has been suggested that the
transcription factors disrupt critical transcriptional programs, such as proliferation,
differentiation and apoptosis, during thymocyte development, promoting T-cell
survival (reviewed in Armstrong and Look, 2005; reviewed in Grabher et al, 2006).
More recently, activating mutations occurring in developmentally important genes have
been identified (Weng et al, 2004; Paietta et al, 2004). The most striking was the
discovery of the presence of activating NOTCH1 mutations in over 50% of T-ALL
patients (discussed in detail below). Additionally, in a smaller percentage of patients
with T-ALL, activating internal tandem duplications or point mutations in the FMS-like
tyrosine kinase 3 (FLT3) gene have also been reported (Paietta et al, 2004;
Van Vlierberghe et al, 2005). FLT3 encodes a receptor tyrosine kinase, which is
important in haematopoietic stem cell development (reviewed in Gilliland and Griffin,
2002). One study reported that these mutations were only present in adult T-ALL cases
with an immature phenotype (CD117/KIT+). Although this was not the case in
paediatric T-ALL patient specimens (Vlierberghe et al, 2005), in both adult and
paediatric T–ALL patient specimens harbouring FLT3 mutations, LYL1 and LMO2
were highly expressed (Paietta et al, 2004; Vlierberghe et al, 2005). This finding may
prove important from a therapeutic perspective, since FLT3 mutations are the most
common mutations associated with AML and are already being targeted in current
clinical trials for AML. Therefore, this class of inhibitor may also be incorporated in
future trials for T-ALL.
31
1.6.3 Oncogene activation in T-ALL
Gene expression profiling of a panel of paediatric T-ALL samples demonstrated
molecularly distinct subgroups which were characterised by the expression of the
transcription factors listed above and the MLL-ENL fusion gene (Ferrando et al, 2002).
Interestingly, the gene expression signatures of the distinct T-ALL subgroups were
noted to correlate with the immunophenotypic features of normal developing
thymocytes arrested at the different stages of differentiation. The molecular subtype of
T-ALL expressing LYL1 was characterised by the expression of genes arrested at the
CD4/CD8 DN stage of T-cell development. On the other hand, cases expressing
HOX11 and HOX11L2 demonstrated the CD4/CD8 DP, early cortical thymocyte stage.
Genes representing the DP stage also characterised TAL1 expressing cases, but in
contrast to the HOX genes, this was typical of the late cortical stage of thymocyte
differentiation, which expresses CD3 and TCRα/β. Subsequently, up-regulation of
TCRγ and TCRδ was found to define cases of T-ALL represented by the MLL-ENL
fusion, consistent with the TCRγδ stage of thymocyte development
(Ferrando et al, 2003). Likewise, T-lymphoblasts with the CALM-AF10 fusion have
also been found to be limited to the TCRγδ stage of thymocyte development or
alternatively to the immature stage with no TCR expression (Asnafi et al, 2003). These
observations underscore the strong relationship between T-cell leukaemogenesis and
normal T-lymphocyte development (reviewed in Armstrong and Look, 2005).
Importantly, several of the transcription factors defining these distinct molecular
subgroups, including HOX11, HOX11L2, TAL1 and LYL1 appear to have prognostic
significance in T-ALL (Ferrando et al, 2002) (Table 1.2), as described in detail below.
32
Table 1.2 Frequency and prognostic significance of molecular abnormalities found
in T-ALL. It is noteworthy that for almost all of the molecular abnormalities found in
T-ALL no consistent prognostic significance has been found. Indeed, despite the
multitude of studies performed, no molecular markers are in clinical use for
prognostication. Validation of molecular markers in large prospective clinical trials is
required.
Molecular abnormality Frequency (%)
Prognostic significance
Comments Reference
HOX11 high level overexpression low level overexpression
9 range 5-20
14
range 4 – 29
Generally favourable
Neutral
One study found that prognosis was treatment dependent
1,2,3,4,5,6,
HOX11L2 overexpression
20
range 10 - 24
Variable
Prognosis appears to depend upon treatment strategy
1,2,4,5,7,8
TAL1 Overexpression
27
range 12 - 62
Variable
2 studies trend for favourable prognosis & 1 adverse
1,2,4,5,7,
8,9,10
CALM-AF10 translocation
7
range 4 – 13
Adverse
Adverse prognosis may be restricted to patients with an immature phenotype
5,7,8,11
MLL-ENL translocation
6
range 4 – 7.6
Favourable
In pre-B ALL, MLL gene rearrangements confer poor prognosis
12,13,14
LYL overexpression
29
range 22 - 35
Adverse
Prognosis based on findings of a single study
1,5
CDKN2A locus deletion (homozygous)
61
30 - 77
Variable
3 studies adverse prognosis & 2 studies neutral
15,16,17,18,
19,20
NOTCH1 activating mutations
51
range 38 - 57
Variable
1 study trend for favourable prognosis, 1 study adverse & 1 study neutral
8,21,22,23
References: 1 Ferrando et al, 2002; 2 Ballerini et al, 2002; 3 Kees et al, 2003a; 4 Cavé et al, 2004; 5 Asnafi et al, 2004; 6 Bergeron et al, 2007; 7 van Grotel et al, 2006; 8 van Grotel et al, 2008; 9 Bash et al, 1993; 10 Bash et al, 1995; 11 Asnafi et al, 2003; 12 Rubnitz et al, 1999; 13 Moorman et al, 2002; 14 Ferrando et al, 2003; 15 Cayuela et al, 1996; 16 Kees et al, 1997; 17 Heyman et al, 1996; 18 Ramakers-van Woerden et al, 2001; 19 Takeuchi et al, 1995; 20 Rubnitz et al, 1997; 21 Weng et al, 2004; 22 Breit et al, 2006; Zhu et al, 2006
33
1.6.3.1 Homeobox genes: HOX11, HOX11L2 and HOXA-D cluster
1.6.3.1.1 HOX11 and HOX11L2
HOX11 is a transcription factor encoding a homeodomain oncoprotein that binds to
DNA (Dube et al, 1991; Hatano et al, 1991; Kennedy et al, 1991; Lu et al, 1991).
HOX11 is aberrantly expressed in approximately 7% of childhood T-ALL patients as a
result of either of two chromosomal translocations, t(7:10) or t(10:14), which place the
HOX11 coding sequence under the transcriptional control of TCRδ /TCRα genes
(Dube et al, 1991; Hatano et al, 1991; Kennedy et al, 1991; Lu et al, 1991).
Overexpression occurs exclusively in T-ALL (Salvati et al, 1995; Kees et al, 2003a)
and frequently in the absence of any translocation (Ferrando et al, 2002;
Kees et al, 2003a). In a recent study almost all patients with T-ALL who expressed
high levels of HOX11 had associated 10q24 chromosomal abnormalities
(Bergeron et al, 2007). High HOX11 overexpression has been reported in
approximately 9% (range 5 to 20%) of T-ALL patients (Table 1.2). On the other hand,
low HOX11 overexpression which is observed in approximately 14% (range 4 to 29%)
of T-ALL patients (Table 1.2), was not associated with any chromosomal
abnormalities, either by conventional karyotyping or FISH, suggesting that this group
may not trigger oncogenic pathways (Bergeron et al, 2007).
The mechanism by which HOX11 exerts its leukaemogenic effect is not yet clear.
Sub-lethally irradiated mice transfused with BM cells overexpressing HOX11 develop
T-ALL like malignancies, but with a very long latency, implying that other mutations
are required (Hawley et al, 1997).
The aberrant expression of HOX11 is generally considered to confer a favourable
prognosis in both paediatric and adult T-ALL patients treated on contemporary
intensive protocols (Ferrando et al, 2002; 2004b; Kees et al, 2003a; Cavé et al, 2004;
34
Bergeron et al, 2007) (Table 1.2). Gene expression profiling, revealed that
HOX11-positive cells were associated with up-regulation of several genes associated
with cellular proliferation in combination with the down-regulation of major anti-
apoptotic genes (BCL2 and BCLXL), thus potentially conferring increased susceptibility
to chemotherapy (Ferrando et al, 2002).
HOX11L2 (HOX11-Like 2) is structurally related to HOX11 and is critical for the
development of central respiratory structures (Shirasawa et al, 2000). HOX11L2 is
aberrantly expressed in approximately 20% of paediatric T-ALL cases (Table 1.2). It
usually arises as a result of a cryptic translocation t(5;14)(q35;q32), which places
HOX11L2 next to the gene BCL11B (Bernard et al, 2001). Less frequently, HOX11L2
is activated by translocations t(5;14)(q34;q11) (Hansen-Hagge et al, 2002) or
t(5;7)(q35;q21) (Su et al, 2004) via juxtaposition to TRCα/δ or CDK6 genes
respectively. Microarray analysis has revealed clustering of HOX11L2-positive samples
with HOX11-positive samples (Ferrando et al, 2002), underscoring the marked
similarity in gene expression profiles associated with HOX11 and HOX11L2. However,
notable differences in several genes involved in signal transduction and chromatin
related genes were observed. The prognostic relevance of these genes is one of the
subjects of this thesis (Chapter 2).
1.6.3.1.2 HOXA-D cluster
The HOXA-D cluster of genes contains a multitude of transcription factors that regulate
anteroposterior tissue patterning during development (reviewed in Kmita and Duboule,
2003) and stem cell self-renewal (reviewed in Pardal et al, 2003; reviewed in Huntly
and Gilliland, 2005a). Approximately 3 to 5% of specimens from T-ALL patients have
been found to contain the chromosomal inversion inv (7) (p15q34) or translocation
t(7;7)(p15;q34), which results in juxtaposition of the TCRβ locus on 7q34-35 to the
35
HOXA cluster on 7p15, culminating in ectopic expression of HOXA10 and HOXA11
(Soulier et al, 2005; Speleman et al, 2005; Cauwelier et al, 2007). A HOXA-TCRδ
translocation has also been described, which interestingly occurred in a patient with a
concomitant CALM-AF10 fusion. This translocation also resulted in deregulated
expression of HOXA cluster genes (Bergeron et al, 2006). The prognostic significance
of these molecular abnormalities has not yet been reported.
1.6.3.2 Helix-loop-helix transcription factors: TAL1 (SCL), TAL2, LYL1, bHLHB1
1.6.3.2.1 TAL1 (SCL), TAL2, bHLHB1
The gene TAL1 (SCL) located on chromosome 1p32, normally functions as a critical
regulator of haematopoiesis, but is not expressed in developing thymocytes
(Robb et al, 1995; Shivdasani et al, 1995). TAL1 is aberrantly activated in T-ALL by
chromosomal translocations t(1;14)(p33;q11) and t(1;14)(p32;q11) which juxtapose
TAL1 next to the TCRδ gene on chromosome 14, and accounts for approximately 1 to
3% of patients (Carroll et al, 1990; Begley et al, 1989). More commonly, in 10 to 26%
of cases, TAL1 is aberrantly activated via deletion of a portion of chromosome 1 which
results in TAL1 coming under the influence of SIL (SCL interrupting locus) forming a
SIL-TAL fusion transcript with TAL1 disruption (Bash et al, 1993; Bernard et al, 1991;
Aplan et al, 1992). Other mechanisms of aberrant gene activation must exist
(Ferrando et al, 2004a), since TAL1 overexpression has been observed in 49% of
children with T-ALL, in one study (Ferrando et al, 2002) and 62% in another
independant study (Bash et al, 1995) (Table 1.2). TAL1 overexpression occurs more
commonly in older children and patients with higher WBC counts at diagnosis. On the
other hand, overexpression of TAL2, a gene highly homologous to TAL1, occurring
through the translocation t(7;9)(q34;32), which juxtaposes TAL2 next to TCRβ is only
rarely detected in T-ALL (Xia et al, 1991). Similarly, the gene bHLH1, a gene
36
juxtaposed next to TCRα by the translocation t(14;21)(q11;q22), is also rarely observed
(Wang et al, 2000).
The precise mechanism by which TAL1 results in leukaemogenesis remains elusive.
Transgenic mouse models revealed leukaemia formation when Tal1 was aberrantly
expressed, albeit with a long latency and relatively low penetrance (Aplan et al, 1997).
It is postulated that TAL1 acts in a dominant-negative fashion by inhibiting the activity
of the transcription factors E2A and HEB. This theory is supported by the observation
that mice lacking E2A develop T-ALL (Bain et al, 1997; Yan et al, 1997) and that
Sil-Tal1 transgenic mice missing the Tal1 transactivation domain develop T-cell
neoplasms (Aplan et al, 1997; O’Neil et al, 2001). Additional evidence is derived from
data revealing that Tal1 transgemic mice on an E2A or HEB heterozygous background
develop T-ALL with increased penetrance and reduced latency, an effect found to be
mediated by repression of the mSin3A/HDAC1 corepressor complex
(O’Neil et al, 2004). TAL1 is also known to collaborate with the LIM domain proteins
LMO1 and LMO2, since Tal1 transgenic mice which overexpress Lmo1 and Lmo2
generate T-cell malignancies with reduced latency (Aplan et al, 1997;
Larson et al, 1996).
More recent data using tamoxifen inducible Tal1 transgenic mice identified TAL1
mediated activation of the NOTCH1 pathway, providing evidence for a functional
relationship between TAL1 and NOTCH1 during T-cell leukaemogenesis
(Göthert et al, 2007). Reports on the prognostic significance of TAL1 have also been
variable, with some investigators reporting a trend towards a favourable prognosis
(Bash et al, 1993; Cavé et al, 2004), whilst others revealed an adverse outcome
associated with TAL1 overexpression (Ferrando et al, 2002).
37
1.6.3.2.2 LYL1
LYL1 (lymphoblastic leukaemia derived sequence 1) is located on chromosome
19p13.2 and is activated in T-ALL by juxtaposition to the TCRβ locus as a result of the
rare translocation t(7;19) (q34;p13) (Mellentin et al, 1989). However, LYL1 is
aberrantly expressed in approximately 22% to 35% of T-ALL cases in the absence of
any detectable translocations (Ferrando et al, 2002; Asnafi et al, 2004) (Table 1.2).
LYL1 tends to be co-expressed in conjunction with LMO2 (see below). Overexpression
of LYL1 in T-ALL blasts has been associated with an unfavourable prognosis, which
was attributed to the up-regulation of several anti-apoptotic genes in this subgroup of
T-ALL patients (Ferrando et al, 2002).
1.6.3.3 LIM domain only zinc finger encoding genes: LMO1 and LMO2
LMO1 (LIM-domain only) and LMO2 occur in T-ALL as a result of translocations
t(11;14)(p15;q11) and t(11;14)(p13;q11) respectively, which juxtapose these genes next
to the TCRα/δ locus (McGuire et al, 1989; Royer-Pokora et al, 1991; reviewed in
Rabbitts, 1998). LMO1 and LMO2 can also be overexpressed in the absence of any
chromosomal translocations and this has been detected in around 45% of patients with
T-ALL (Ferrando et al, 2002). The oncogenic capacity of LMO1 and LMO2 has been
demonstrated in transgenic mouse models (Fisch et al, 1992; McGuire et al, 1992,
Larson et al, 1994; Neale et al, 1995). Interestingly, in patients with T-ALL, LMO2
expression often occurs in conjunction with either TAL1 or LYL1, whereas LMO1 is
frequently associated with TAL1 expression only, indicating a common oncogenic
pathway (Ferrando et al, 2002). As already described above, this relationship is
supported by transgenic mouse models (Aplan et al, 1997; Larson et al, 1996).
38
1.6.3.4 MLL-ENL
MLL gene rearrangements are the hallmark of pre-B ALL and AML in infants.
However, they are also found in approximately 4 to 8% of patients with T-ALL
(Ferrando et al, 2002; Moorman et al, 2002), which typically fuses the MLL gene to the
ENL gene, as a result of the translocation t(11;19) (q23;p13.3). Pre-B ALL with an
MLL rearrangement are characterised by up-regulation of several HOXA-D cluster
genes, including HOXA9, HOXA10, HOXC6 and the cofactor MEIS1
(Armstrong et al, 2002). Similarly, gene expression profiling has identified
up-regulation of these genes in MLL-ENL T-ALL cases (Ferrando et al, 2003;
Soulier et al, 2005), implicating these genes in T-cell leukaemogenesis in T-ALL cases
with MLL gene rearrangements (Ferrando et al, 2003; Soulier et al, 2005). In infants,
MLL gene rearrangements are associated with a particularly poor prognosis
(reviewed in Pui and Campana, 2007b). Paradoxically, the MLL-ENL fusion in T-ALL
has been associated with a favourable prognosis (Rubnitz et al, 1999).
1.6.3.5 CALM-AF10
This fusion has been reported in 4 to 13% of childhood cases of T-ALL
(Asnafi et al, 2003; 2004; van Grotel et al, 2006; 2008) (Table 1.2). It occurs as a result
of the cryptic translocation t(10;11)(p13;q14-21), which fuses the CALM (Clathrin
Assembly protein-like Lymphoid-Myeloid Leukaemia) gene to AF10. Interestingly,
MLL is also infrequently fused to AF10 (Dreyling et al, 1998). Analogous to specimens
from T-ALL patients with the MLL-ENL fusion, gene expression profiling has also
demonstrated up-regulation of several HOXA cluster genes, including HOXA5, HOXA9
and HOXA10 as well as MEIS1 in specimens from patients with the CALM-AF10
fusion (Soulier et al, 2005; Dik et al, 2005; Bergeron et al, 2006), suggesting that
39
CALM-AF10 and MLL-ENL activate common oncogenic pathways (Graux et al, 2006).
However, in contrast to MLL-ENL, the presence of the CALM-AF10 fusion has been
associated with a poor prognosis for T-ALL patients (Asnafi et al, 2003;
van Grotel et al, 2006; 2008). Notably, the prognostic significance of the CALM-AF10
fusion appears to depend upon the maturational stage, since only T-lymphoblasts with
the CALM-AF10 fusion and an immature stage were associated with an adverse
outcome (Asnafi et al, 2003).
1.6.3.6 NUP214-ABL1
The novel cryptic ABL1 gene amplification has been observed in specimens from
approximately 3 to 6% of children with T-ALL (Graux et al, 2004; Barber et al, 2004).
It occurs as a consequence of episomal fusion between the ABL1 gene and the NUP214
gene, producing a NUP214-ABL1 fusion gene (Graux et al, 2004). Expression profiling
revealed that this subtype of T-ALL is associated with overexpression of HOX11 or
HOX11L2 and deletion of the CDKN2A locus (Graux et al, 2004). Alternative ABL1
fusion partners in T-ALL include BCR (BCR-ABL1), ETV (ETV-ABL1) and EML1
(EML1-ABL1), but these are only occasionally identified (<1%)
(reviewed in Graux et al, 2006). Interestingly, Ballerini et al, (2005) reported a case of
a child diagnosed with T-ALL who failed induction and died one month after
diagnosis. The patient’s T-lymphblasts were found to overexpress HOX11L2 due to the
presence of the translocation t(5;14) (q35;q14), as well as harbour amplification of the
NUP214-ABL1 fusion in the long arm of one chromosome 2. The authors proposed that
the presence of the NUP214-ABL1 fusion may explain the variable outcomes observed
for T-ALL patients whose T-lymphoblasts express HOX11L2 (Ballerini et al, 2005).
However, the prognostic significance of the NUP214-ABL1 fusion in paediatric T-ALL
remains to be determined. In adults with T-ALL, the NUP214-ABL1 fusion was not
40
associated with patient outcome (Burmeister et al, 2006). Importantly, the abnormal
tyrosine kinase produced by these fusions appears to be sensitive to imatinib in vitro
(Graux et al, 2004; De Keersmaecker et al, 2006), providing a novel therapeutic
approach for this subtype of T-ALL.
1.6.3.7 MYB
The transcription factor MYB, which normally functions as an important regulator of
haematopoiesis, has recently been identified as another TCR partner gene in T-ALL
patients. The MYB gene has been found juxataposed next to TCRβ as a result of the
chromosomal translocation t(6;7)(q23;q34) in approximately 7% of specimens from
T-ALL patients (Clappier et al, 2007). Additionally, using array comparative genomic
hybridisation (array-CGH), duplications have also been identified in 8 to 15% of cases,
as an alternative mechanism of gene overexpression (Clappier et al, 2007;
Lahortiga et al, 2007). Notably, knockdown of MYB using short interfering RNA
(siRNA) in T-ALL cell lines harbouring NOTCH1 mutations, in conjunction with a
gamma-secretase inhibitor resulted in marked synergism in cytotoxicity, identifying
MYB as a potential novel therapeutic target in T-ALL (Lahortiga et al, 2007).
1.6.3.8 NOTCH signalling in T-ALL
NOTCH signalling regulates numerous important cellular functions, including cell fate
decisions, differentiation, proliferation and apoptosis during development in a variety
of tissue types (reviewed in Artavanis-Tsakonas et al, 1999). Notably, NOTCH1
regulates multiple haematopoietic lineage decisions, including the establishment of the
earliest T-cell progenitors (reviewed in Pear and Aster, 2004). NOTCH1 was first
associated with human cancer by the discovery of the rare chromosomal translocation
t(7;9) which occurs in approximately 1% of all T-ALL cases (Ellisen et al, 1990). This
41
translocation juxtaposes NOTCH1 adjacent to the TCRβ locus, resulting in the aberrant
expression of a truncated active form of NOTCH1. Evidence directly supporting a
leukaemogenic role for NOTCH1 signalling in T-ALL comes from mouse models.
Sub-lethally irradiated mice transfused with BM cells transduced with a truncated
activated form of Notch1 developed T-ALL (Pear et al, 1996). Additionally, the
aberrant expression of the NOTCH family members, Notch2 and Notch3, also results in
the development of T-ALL/lymphoma (reviewed in Pear and Aster, 2004). Deregulated
NOTCH signalling has also been implicated in numerous other cancers including,
pancreatic, breast, prostate and CNS tumours (reviewed in Sjölund et al, 2005).
Recently, NOTCH1 signalling has emerged as a major player in T-cell
leukaemogenesis, following the discovery that activating NOTCH1 mutations were
present in 56% of paediatric patients with T-ALL (Weng et al, 2004). The mutations
were found to occur either in the heterodimer domain, leading to ligand independent
activation, in the PEST domain, resulting in increased stability of NOTCH1
intracellular domain (ICN1) in the nucleus or in combination. Analysis of specimens
obtained from pre-B ALL revealed no NOTCH1 mutations, suggesting that these
mutations are restricted to T-ALL. Notably, these mutations were present in most of the
molecular subtypes which characterise T-ALL, implying that NOTCH1 may
co-operate with other proto-oncogenes in T-cell leukaemogenesis (reviewed in
Armstrong and Look, 2005; reviewed in Grabher et al, 2006).
Recent studies have identified downstream targets of NOTCH1 signalling, including
the oncogene MYC (Palomero et al, 2006; Weng et al, 2006; Sharma et al, 2006) and
the NFκB pathway (Vilimas et al, 2007), thus beginning to unravel the mechanisms
responsible for NOTCH mediated T-cell leukaemogenesis. Studies investigating the
prognostic significance of activating NOTCH1 mutations have yielded conflicting
results (Breit et al, 2006; Zhu et al, 2006; van Grotel et al, 2008), thus additional
42
studies are required to further assess the prognostic significance of NOTCH1 activating
mutations.
1.6.4 Tumour suppressor gene silencing in T-ALL
In addition to oncogene activation, simultaneous deletion of tumour suppressor genes is
common in many specimens from T-ALL patients. The gene TP53 is the single most
commonly inactivated tumour suppressor in cancer (Olivier et al, 2002), yet, it is only
rarely mutated in T-ALL, except at relapse, where 30% of samples have been
demonstrated to harbour TP53 alterations (reviewed in Imamura et al, 1994). In
contrast, inactivation of the CDKN2A (cyclin-dependent kinase inhibitor 2A) locus on
chromosome 9p21, which encodes two potent but distinct tumour suppressor proteins,
p16INK4a and p14ARF, is the most frequent genetic abnormality in T-ALL, present in the
majority of patients (reviewed in Drexler, 1998) (Table 1.2). The CDKN2A locus is the
primary target of deletions involving 9p21. The CDKN2A locus encodes for p16INK4a
and p14ARF proteins due to the presence of distinct exon structures, which result in
alternative reading frames during translation (Quelle et al, 1995). Both p16INK4a and
p14ARF are involved in cell cycle regulation, but act through different pathways. The
protein p16INK4a inactivates the RB1 (Retinoblastoma 1) tumour suppressor by
inhibiting the cyclin D-Cdk4/6 complexes which normally phosphorylate it
(reviewed in Roussel, 1999), whereas p14ARF activates TP53 by inhibiting HDM2
which normally degrades TP53 (reviewed in Sharpless and DePinho, 1999).
Accordingly, inactivation of the CDKN2A locus effectively results in loss of the TP53
and the RB1 tumour suppresor pathways (Mullighan et al, 2005). Frequently, the
adjacent locus CDKN2B, which encodes another tumour suppressor gene, p15INK4B, is
concomitantly deleted (Drexler, 1998; Carter et al, 2001). Deletion of the CDKN2A
locus was particularly associated with the HOX11 and TAL1 molecular subgroups of
43
T-ALL (Ferrando et al, 2002). Alternative mechanisms of suppressing the CDKN2A
locus include mutation (Okamoto et al, 1994), promoter hypermethylation
(Merlo et al, 1995; Herman et al, 1996) and overexpression of the CDKN2A inhibitor,
BMI1. Overexpression of BMI1 has been identified exclusively in T-ALL patients
harbouring the CALM-AF10 fusion, which notably, has not been shown to possess
deletions in the CDKN2A locus (Dik et al, 2005). Whilst another group, identified
BMI1 overexpression in association with both the CALM-AF10 fusion and MLL-related
cases (Soulier et al, 2005). Therefore, inactivation of the CDKN2A/B loci occurs in
almost all T-ALL patients. Homozygous deletions of the CDKN2A locus have been
associated with an inferior prognosis (Heyman et al, 1996; Kees et al, 1997;
Ramakers-van Woerden et al, 2001). Another study which analysed both T-ALL and
pre-B ALL patients found that hemizygous deletion of the CDKN2A locus also
conferred a poor prognosis (Carter et al, 2001). However, other studies found no
prognostic value to such deletions (Takeuchi et al, 1995; Rubnitz et al, 1997)
(Table 1.2).
Deletions involving the long arm of chromosome 6 are the single most common
chromosomal abnormality in T-ALL, detected in approximately 20 to 30% of cases
(Heerema et al, 1998; Schneider et al 2000) and occurring more commonly in T-ALL
than pre-B ALL (Heerema et al, 2000). Molecular studies have identified a common
region of loss of heterozygocity (LOH) between 6q16 and 6q21 (Merup et al, 1998;
Takeuchi et al, 1998). However, the precise identification of a tumour suppressor gene
in that region remains elusive (Sinclair et al, 2004). No prognostic value appears to be
associated with the presence of 6q deletions (Heerema et al, 1998; 2000;
Schneider et al, 2000).
Recently, Wolfrain et al (2004) showed that loss or reduction of SMAD3 protein, a
member of the transforming growth factor-β (TGF-β) signal transduction pathway, was
44
found exclusively in paediatric T-ALL specimens. In mice, a reduction in Smad3 in
association with loss of p27kip1 resulted in T-cell leukaemogenesis, strongly implicating
SMAD3 as a novel tumour suppressor in paediatric T-ALL. The mechanisms
responsible for loss or reductions of SMAD3 protein expression remain unknown.
The gene FBW7 encodes for a protein ubiquitin ligase on chromosome 4q31.3, which
binds and degrades ICN1, MYC and cyclin E. Inactivating mutations in this gene have
recently been identified in 9 to 16% of specimens from patients with T-ALL
(O’Neil et al, 2007; Thompson et al, 2007). In T-ALL cell lines, mutated FBW7 is
unable to bind to NOTCH, resulting in elevated ICN1 levels (O’Neil et al, 2007;
Thompson et al, 2007). Furthermore, mutated FBW7 is unable to degrade MYC
(O’Neil et al, 2007). These data indicate that FBW7 is a novel tumour suppressor gene
in T-ALL, which normally acts by keeping the NOTCH pathway in check
(O’Neil et al, 2007; Thompson et al, 2007). Notably, T-ALL cell lines harbouring such
mutations were found to be resistant to gamma-secretase inhibitors, suggesting that loss
of this gene is a potential mechanism of drug resistance in T-ALL
(Thompson et al, 2007).
In summary, it is telling that the CDKN2A tumour suppressor locus is inactivated in
leukaemia cells of almost all T-ALL patients, suggesting that T-cell leukaemogenesis
occurs via multistep oncogenic pathways in co-operation with the loss of tumour
suppressor genes (reviewed in Armstrong and Look, 2005;
reviewed in Grabher et al, 2006; reviewed in Graux et al, 2006).
45
1.7 Future risk-stratification
1.7.1 The genomic era
Genomic techniques, including gene expression arrays and single-nucleotide
polymorphism (SNP) arrays have revolutionised molecular biology, affording
researchers the ability to perform high density molecular profiling of tumour cells.
Gene expression microarrays are available in several distinct platforms including
commercially available pre-fabricated oligonucleotide arrays from Affymetrix
(Affymetrix GeneChips) and Agilent (Agilent 60-mer oligonucleotide microarrays) or
in-house custom arrays including spotted cDNA arrays and spotted oligonucleotide
arrays (reviewed in Harrington et al, 2000). Microarray experiments performed as part
of this thesis were carried out using Affymetrix GeneChips. This platform contains
thousands of 25-mer oligonucleotides (20 per gene), which uniquely identify transcripts
of thousands of genes within the human genome. They are fabricated using
photolithographic chemistry on silicon wafers. Extracted RNA is labelled with a
fluorescent dye and subsequently hybridised to gene-specific probes located on the
array (reviewed in Quackenbush, 2006). To improve accuracy and reproducibility each
gene is represented by multiple probes, which are located throughout the array,
comprising a probe set. The arrays are scanned using a confocal laser which produces a
fluorescent signal. This signal is subsequently converted into an expression level for
each gene; the higher the fluorescence signal, the larger the level of gene expression.
The raw data generated requires processing before analysis. The first step undertaken is
data normalisation, a technique performed to adjust for variation in technical
differences between GeneChips (reviewed in Quackenbush, 2006). After normalisation
a variance filter is generally applied, to remove non-informative probe sets
(reviewed in Quackenbush, 2006). Notably, different investigators apply diverse
46
filtering criteria. In this thesis filtering criteria, as described in Hoffmann et al (2005),
were applied to the normalised array data. Following normalisation and filtering, data
analysis can commence. Analysis methods are broadly divided into two categories;
unsupervised and supervised (reviewed in Simon et al, 2003; reviewed in
Quackenbush, 2006). Unsupervised analysis is an unbiased technique, where no
information about the samples is used. This is generally performed at the outset, to look
for unbiased relationships and trends in the data. Unsupervised analysis methods cluster
data according to similarities in the gene expression profiles between samples. The
most commonly applied technique to perform unsupervised clustering is hierarchical
clustering (reviewed in Miller et al, 2002). Other clustering tools include k-means
clustering and self-organising maps (reviewed in Miller et al, 2002;
reviewed in Quackenbush, 2006). To identify gene signatures related to a specific
feature, such as relapse, supervised analysis is required. Various different supervised
learning algorithms have been utilised, such as support vector machines, artificial
neural networks and k-nearest neighbours (reviewed in Miller et al, 2002). These
algorithms have been employed to analyse the same data set (Shipp et al, 2002;
Yeoh et al, 2002; Ross et al, 2003) and these studies have demonstrated that these
different learning algorithms performed equally well and yielded very similar results.
As part of this thesis however, array data was analysed using alternative techniques,
known as robust multi-array analysis (RMA) in combination with the Random Forrest
(RF) supervised learning algorithm. RMA is a probe-level data extraction algorithm
which has been shown to provide greater sensitivity and specificity in detection of
differential gene expression compared to other expression measures,
as well as significantly less variable and more reliable gene expression values
(Irizarry et al, 2003a; Irizarry et al, 2003b). RF is a supervised decision-tree based
approach which contains a built-in reiterative process, which combines bootstrap
47
sampling via bagging and random feature selection. Consequently, in contrast to many
other analytical methods, this approach does not require that a portion of the samples be
removed from the analysis for later validation. Since most microarray experiments are
based on a relatively small number of samples and a much larger number of data
points, cross-validation procedures that require a considerable portion of the samples to
be removed as a validation set can result in decreased prediction certainty and data over
fitting. Indeed, data over fitting is one of the most challenging problems associated with
microarray data analysis (Michiels et al, 2005). The use of the RF approach as a
supervised learning algorithm has been previously verified (Zhang et al, 2003). The
combination of RMA and RF has also been previously validated as an alternative tool
for the analysis of oligonucleotide arrays (Hoffmann et al, 2006). These analytical tools
were found to yield higher accuracies when compared to some of the other methods
listed above (Hoffmann et al, 2006). Techniques to reduce data over fitting have been
proposed, such as complete cross-validation, where all the steps involved in the
construction of a predictor are cross-validated (Simon et al, 2003) and repeated random
sampling (Michiels et al, 2005). These approaches are especially useful when sample
numbers are limited. However, the ultimate test is to validate candidate genes in a
completely independent set of validation samples, which were not involved in the
selection of candidate predictor genes (Simon et al, 2003).
In recent years gene expression profiling has been widely applied, resulting in novel
insights into the biology of ALL. Investigators have used this technology to identify
signature gene expression profiles for specific molecular subgroups of ALL (Yeoh et
al, 2002; Ross et al, 2003), uncover novel molecular subtypes of ALL
(Yeoh et al, 2002) and examine the mechanisms governing leukaemia relapse
(Beesley et al, 2005; Lugthart et al, 2005; Bhojwani et al, 2006). In addition, this
technology has been applied to develop outcome classifiers to improve prognostication
48
(Yeoh et al, 2002; Chiaretti et al, 2004), response to therapy predictions
(Cairo et al, 2005) and identify genes associated with drug resistance
(Holleman et al, 2004; 2006; Winter et al, 2007). This technology has resulted in the
characterisation of the molecular genetic changes which occur in T-ALL, leading to
important discoveries regarding T-cell leukaemogenesis (Ferrando et al, 2002). More
recently, gene expression profiling was used to elucidate the molecular basis of
resistance to novel targeted therapies for T-ALL (Palomero et al, 2007). In this way,
genetic profiling can thus be both diagnostic and prognostic, and the use of genomic
technologies is expected to become commonplace in the clinic to refine patient
classification and stratification at diagnosis (Figure 1.2). More recently, genome-wide
analysis of ALL using SNP arrays has been used to uncover novel molecular alterations
leading to the disruption of key pathways involved in leukaemogenesis
(Mulligan et al, 2007).
The utilisation of gene expression profiling to uncover novel predictive markers for
childhood T-ALL is a major subject of this thesis (Chapters 3 and 4). Furthermore,
genes associated with a signature gene expression profile of relapse provide potential
targets for novel therapies, which is the focus of Chapter 5.
1.7.2 Identification of predictive markers
Although the outcome for childhood T-ALL has improved markedly over the last forty
years, in the past decade survival rates have been almost static at around 75%. Without
refined patient stratification, further empirical changes in treatment and/or additional
treatment intensification are unlikely to result in significant improvements in outcome.
In fact, further treatment intensification in unselected groups is more likely to result in
increased morbidity and mortality, without significant gains in survival. Improvements
in molecular techniques now allow the more accurate assessment of remission by
49
measuring MRD, permitting more sophisticated treatment stratification. Most
cooperative groups have included the measurement of MRD to guide treatment in their
current up-front ALL studies.
The ultimate goal is to develop individualised treatment strategies based on the
underlying biology of patient’s leukaemic blasts as well as the patient’s individual
pharmacogenetic and pharmacodynamic characteristics. Improved understanding of
T-lymphoblast biology will permit more precise disease risk classification and
improved tailoring of therapy. Moreover, targeted disruption of molecular alterations
critical for T-cell leukaemogenesis and/or T-lymphoblast survival, using specific small
molecule inhibitors, is expected to improve patient outcomes whilst reducing toxicity.
50
Figure 1.2 A proposed schema for selecting the optimal therapy for children with
T-ALL at presentation. To assist in selecting the most appropriate therapy, data from
molecular analysis of blast cells are used in conjunction with clinical information to
determine an individual patient’s risk category more accurately. Therapy can be further
modified based on a patient’s response to therapy, using minimal residual disease
(MRD) analysis (adapted from Pui and Evans, 2006a and Gilbertson, 2004).
Pharmacogenomics
Constitutional DNA
T-ALLblasts
Diagnosis
Discovery of pathways critical for T-cell leukaemogenesis& maintenance
Novel therapeutic targets
Refinement of stratification based on patient response to therapy (MRD analysis)
Highrisk
Novel/experimental
therapies
Molecular stratification
Standard risk
Standard therapy
compared to experimental
arm
Lowrisk
Strategiesto reducetherapy
RNA DNA ProteinProtein
51
1.8 Thesis hypothesis and objectives
The hypothesis of this thesis is that gene expression profiling could uncover a signature
linked to outcome for paediatric patients with T-ALL, and that genes constituting a
relapse signature would provide rational targets for novel therapies directed at such
genes.
The studies presented as part of this thesis focus on gene expression in childhood
T-ALL. The first part of the thesis further investigates the prognostic significance of
the homeobox genes HOX11/TLX1 and HOX11L2/TLX3. In the second part of this
thesis a genome-wide approach, using DNA oligonucleotide gene expression array
technology, was undertaken to define novel genes associated with outcome; “molecular
signature of relapse”.
The specific aims of this study were to:
1) Investigate the prognostic significance of the transcription factor HOX11L2/TLX3 in
a cohort of paediatric T-ALL patients treated according to CCG protocols
2) Verify the gene expression levels obtained using oligonucleotide microarrays
3) Use gene expression profiling to identify a novel gene expression signature for
relapse in paediatric T-ALL
4) Apply the knowledge gained from genes associated with a signature gene expression
profile for relapse to identify and test a potential novel agent for the treatment of
childhood T-ALL
52
CHAPTER 2
SIGNIFICANCE OF HOX11L2/TLX3 EXPRESSION IN CHILDREN
WITH T-CELL ACUTE LYMPHOBLASTIC LEUKAEMIA ON
CHILDREN’S CANCER GROUP PROTOCOLS
2.1 Abstract
The homeobox gene HOX11/TLX1 appears to confer a good prognosis in childhood
T-cell acute lymphoblastic leukaemia (T-ALL). On the other hand, discrepancy exists
about the prognostic significance of patients whose leukaemia cells express the
structurally related gene, HOX11L2/TLX3. Quantitative real-time RT-PCR was used to
determine the level of expression of HOX11 and HOX11L2 in 40 diagnostic paediatric
T-ALL bone marrow specimens obtained from patients treated on Children’s Cancer
Group (CCG) risk-adjusted protocols. Ten (25%) of the patient specimens expressed
HOX11 compared with seven (18%) for HOX11L2. The estimated 5 year relapse free
survival (RFS) rate for all cases was 64% (s.e. ±8%) and the overall survival rate was
56% (s.e. ±9%). Patients whose T-ALL cells expressed HOX11L2 had an excellent
prognosis (100% 5 year RFS) compared to those not expressing HOX11L2 (54% [s.e. ±
9%]) (P = 0.04). This study revealed that HOX11L2 expression conferred an excellent
prognosis on children with T-ALL treated on CCG risk-adjusted protocols.
53
2.2 Introduction
T-cell acute lymphoblastic leukaemia (T-ALL) accounts for approximately 15% of
cases of childhood ALL. In earlier paediatric ALL clinical trials, the T-cell phenotype
was associated with an unfavourable prognosis. However, the use of modern
risk-adjusted multiagent treatment regimes has been reported to override this
difference. While a 5 year event-free survival rate of approximately 70–75% is
currently achieved in childhood T-ALL, a significant proportion of patients continue to
relapse. Importantly, relapsed T-ALL confers an extremely poor prognosis.
With improvements in the outcome of patients with T-ALL there is a pressing need to
identify prognostic molecular markers to aid patient stratification. This is especially
important since, in contrast to B-lineage ALL, cytogenetic abnormalities in T-ALL
have failed to provide comparable prognostic information (Heerema et al, 1998).
Studies have revealed that certain transcription factors associated with T-ALL might be
useful for prognosis. The most notable examples are HOX11/TLX1 and
HOX11L2/TLX3, two structurally related homeobox genes. HOX11 is aberrantly
expressed in childhood T-ALL as a result of either of two chromosomal translocations,
t(7;10)(q35;q24) and t(10;14)(q24;q11), which place the HOX11 coding sequence
under the transcriptional control of T-cell receptor regulatory elements. HOX11L2 is
involved in a cryptic translocation, t(5;14)(q35;q32), detectable only by FISH
(Bernard et al, 2001). Both paediatric and adult T-ALL patients expressing HOX11 in
their blast cells have been demonstrated to have an excellent prognosis
(Ferrando et al, 2002; 2004b; Kees et al, 2003a). On the other hand, the prognostic
significance of HOX11L2 expression is equivocal. Two studies reported expression of
this gene to be associated with a very poor prognosis (Ferrando et al, 2002;
Ballerini et al, 2002) whereas a larger study revealed no difference in prognosis
(Cavé et al, 2004).
54
2.3 Methods
Quantitative real-time RT-PCR (qRT-PCR) (TaqMan) technology was employed to
determine the level of expression of HOX11L2 and HOX11 in 40 diagnostic paediatric
T-ALL bone marrow specimens obtained from patients treated on Children's Cancer
Group (CCG) risk-adjusted protocols between 1984 and 2002. T-ALL lymphoblasts are
thought to originate from normal T-lymphocyte precursors arrested at various stages of
thymocyte development. Putative normal cellular counterparts comprising normal
CD34+ bone marrow (stem) cells (n=3), normal thymocytes (n=6) and peripheral
T cells (n=6) were included as controls. qRT-PCR analysis was performed and
expression levels of test genes were determined as a ratio to a reference gene, ACTB, as
previously described (Kees et al, 2003a). The reference cell lines used were PER-255
for HOX11 and PER-487 for HOX11L2. For the gene HOX11L2 primers and probe
were obtained from Applied Biosystems (ABI Assays on Demand,
www.appliedbiosystems.com). All samples were tested in duplicate. Kaplan–Meier
survival analysis was conducted on relapse-free survival (RFS) and overall survival
(OS) data. The log-rank test was used for comparison of outcome for patient groups
according to the expression of HOX11L2 and HOX11. χ2 test and Fisher's exact test
were used to assess patient characteristics.
2.4 Results and Discussion
HOX11L2 and HOX11 expression were undetectable in all 15 controls, thus any level
of detection in patient specimens was regarded as positive expression of either HOX11
or HOX11L2. Seven (18%) of the patient specimens expressed HOX11L2 compared
with ten (25%) for HOX11. Only three of the 10 (7.5% of the whole cohort)
HOX11-positive specimens exhibited high expression levels. The average expression
level of the high HOX11 expressors was approximately 550 times higher (mean 0.89
55
(range 0.1–1.99)) than for low HOX11 expressors (mean 0.0016 (range 0.00009–
0.0095)). The significance of low-level expression of HOX11 is not known and most
investigators include only high HOX11 expressors in their analysis (Ferrando et al,
2002; 2004b; Asnafi et al, 2004; Cavé et al, 2004). In contrast, HOX11L2 was
expressed at high levels in all specimens (mean 2.08 (range 0.3–8.74)) (Figure 2.1).
These results are consistent with previously published frequencies since HOX11L2
expression in paediatric T-ALL was reported to range from 10 to 24% (mean 20%)
(Bernard et al, 2001; Ferrando et al, 2002; 2004b; Ballerini et al, 2002; Berger et al,
2003; Asnafi et al, 2004; Cavé et al, 2004;). On the other hand, 6–20% (mean 11%) of
paediatric T-ALL specimens expressed high levels of HOX11 (Ferrando et al, 2002;
Ballerini et al, 2002; Kees et al, 2003a; Berger et al, 2003; Asnafi et al, 2004;
Cavé et al, 2004) compared with 4–29% (mean 13%) for low levels of HOX11
expression (Ferrando et al, 2002; Ballerini et al, 2002; Kees et al, 2003a). In agreement
with other studies, HOX11L2 and high levels of HOX11 expression were found to be
mutually exclusive (Ferrando et al, 2002; Ballerini et al, 2002; Berger et al, 2003;
Asnafi et al, 2004; Cavé et al, 2004). One of our specimens expressed HOX11L2 (level
of 0.3) in combination with a low level (0.0005) of HOX11 (marked by arrow in Figure
2.1). The expression of HOX11L2 in association with low levels of HOX11 has been
previously demonstrated (Asnafi et al, 2004) and is of unknown significance.
Cytogenetic analysis was successful in 65% (26/40) of specimens and of these 58%
(15/26) revealed a chromosomal abnormality. None of the patient samples expressing
HOX11 were found to have a translocation involving t(7;10) or t(10;14), further
supporting the observation that HOX11 deregulation can occur in the absence of these
translocations (Ferrando et al, 2002; 2004b; Ballerini et al, 2002; Kees et al, 2003a;
Cavé et al, 2004). Insufficient material was available for FISH analysis to detect
t(5;14)(q35;q32) in specimens expressing HOX11L2.
56
Figure 2.1 Expression of HOX11L2/ACTB and HOX11/ACTB in 40 paediatric
T-ALL specimens. Only specimens expressing HOX11L2 (n=7) and HOX11 (n=10)
are depicted. A solid black line separates high and low expressors. The arrow denotes
patient specimen that expresses both HOX11L2 and HOX11.
To examine gene expression in leukaemia cells and clinical outcome, we estimated the
5 year RFS rate and the OS rate, which was feasible for 39 patients. The median
follow-up time for nonrelapse patients was 4.5 years (range 1.2–12.8 years). A total of
13 (33%) patients relapsed, with a median time of 2 years (range 7.7 months–4.2
years). One patient failed to achieve remission, with greater than 25% blasts in the bone
marrow (M3) by the end of induction and died a short time later from sepsis. This
patient's blasts did not express HOX11L2 or HOX11. The 5 year RFS rate for all cases
was 64% (s.e.±8%) and the OS rate was 56% (s.e.±9%). Patients whose T-ALL cells
57
expressed HOX11L2 had an excellent prognosis (100% 5 year RFS) compared to those
not expressing HOX11L2 (54% (s.e.±9%)) (P=0.04) (Figure 2.2a) contrasting with
other reports. There was no significant difference in the 5 year RFS rate according to
HOX11 expression in blast cells, for HOX11 high level 33% (s.e.±27%) vs. HOX11 low
level 57% (s.e.±19%) vs. HOX11-negative 69% (s.e.±9%) (P = 0.56) (Figure 2.2b).
Accordingly, the comparison between HOX11-positive (high and low levels) vs.
HOX11-negative revealed no significant difference in the 5 year RFS. The 5 year RFS
was 50% (s.e.±16%) for HOX11-positive vs. 69% (s.e.±9%) for HOX11-negative
patients (P = 0.29). The OS with respect to expression of both genes essentially
mirrored the RFS (Figure 2.2c and d). A statistically significant difference in 5 year
RFS rate for girls compared to boys, 100 vs. 53% (s.e.±9%) (P = 0.03) was noted, a
feature often observed in T-ALL. Sex ratio among HOX11L2 expressors was not
significantly different from that of non-expressors (P = 0.32), suggesting that gender
does not explain the prognostic effect of HOX11L2 expression. There were no
statistically significant associations between the expression of HOX11 or HOX11L2
and white cell count (WCC), age or status of day 7 bone marrow. Furthermore, we
found no significant difference in 5 year RFS rate according to WCC, age or status of
day 7 bone marrow.
58
Figure 2.2 Clinical outcomes for 39 paediatric T-ALL patients according to
expression status of HOX11L2 and HOX11. (a and b) Relapse-free survival. (c and d)
Overall survival. X=censored patients.
This is the first study demonstrating a favourable prognosis associated with the
expression of HOX11L2 in T-ALL blasts. The observed differences in outcome could
be due to the relatively small patient cohort investigated. This may also be the situation
for the two studies demonstrating a poor prognosis associated with HOX11L2
expression (Ferrando et al, 2002; Ballerini et al, 2002) particularly since the
significantly larger study conducted by Cavé et al (2004) reported no difference in the
outcome of paediatric patients whose leukaemia cells expressed HOX11L2.
Alternatively, the prognostic impact of HOX11L2 expression may be dependent upon
59
the therapeutic regimen utilised. Indeed, distinct treatment strategies compared to those
used in this study were employed in the studies reporting a poor prognosis associated
with the aberrant expression of HOX11L2 (Ferrando et al, 2002; Ballerini et al, 2002).
The majority of patients in the present study were treated on CCG-modified BFM
(Berlin-Frankfurt-Münster) protocols. This notion is supported by our previous findings
showing that the good prognosis associated with HOX11 expression was evident only
for patients on a particular treatment protocol, implying that HOX11-positive cells may
be more sensitive to specific therapies compared to HOX11-negative cells
(Kees et al, 2003a). On the other hand, in the study conducted by Cavé et al (2004),
patients were also treated with BFM-based therapy, and no difference was found in the
outcome of patients whose blasts expressed HOX11L2. However, the authors
commented that their data was consistent with the premise that HOX11L2 expression
may be associated with a good prognosis, by virtue of a higher proportion of
HOX11L2-expressing samples displaying CD1a+ (cortical/intermediate)
immunophenotype, that has been associated with a superior prognosis. The most
notable difference in therapy that may account for the improved outcome of the patient
cohort expressing HOX11L2 investigated in this study compared to the two studies
which revealed association with poor prognosis (Ferrando et al, 2002;
Ballerini et al, 2002) is the inclusion of a re-intensification (re-induction) phase, a
critical component of BFM therapy for ALL. A re-intensification phase was not
included in the treatment protocol (FRALLE 93 very high-risk arm) of a significant
proportion of patients (33%) investigated by Ballerini et al (2002) or for two of the
three protocols (St Jude Total Therapy Studies XI and XII) utilised to treat patients
studied by Ferrando et al (2002). In contrast, these treatment protocols applied
alternative therapeutic strategies.
60
The favourable prognosis reported to be associated with HOX11 expression was not
replicated in this study. This may have been due to the size of the study. Using
microarray technology, Ferrando et al (2002), found that HOX11-positive cells
expressed several genes associated with cellular proliferation. Thus, the authors
proposed that the better prognosis for patients with HOX11-positive T-ALL could be
linked to the high proliferation rate, making them more susceptible to the effects of
chemotherapy, which preferentially targets actively dividing cells. Furthermore, these
investigators demonstrated that specimens expressing HOX11L2 had remarkably
similar genetic profiles to HOX11 expressing specimens, except for notable differences
in several genes involved in signal transduction and chromatin related genes associated
with HOX11-positive cells. Considering the similarities between HOX11 and
HOX11L2, it is equally plausible that patients, whose leukaemic blasts express either of
these two genes, should have similar prognosis. Hence, the strikingly different patient
outcome between patients with HOX11 and HOX11L2-positive cells was very
surprising (Ferrando et al, 2002).
While, the relatively small number of patient specimens investigated in this study
precludes any definite conclusions, the marked difference in outcome observed here for
patients with HOX11L2-expressing cells compared to other studies supports the
principle that specific biological subtypes of leukaemia may be more sensitive to
distinct therapeutic regimes. Analysis of a larger cohort of uniformly treated patients is
warranted to verify these results.
61
2.5 Author contributions
NGG designed and performed the research, analysed the data and wrote the manuscript.
PAJ carried out the statistical analysis. HNS provided statistical data and advice. GHR
was involved in the concept for the study and he contributed many of the specimens
and clinical data. DLB and URK supervised all aspects of the study and preparation of
the manuscript.
2.6 Acknowledgements
This work was supported by the National Childhood Cancer Foundation Laura and
Greg Norman Fellowship (NGG), the Children’s Leukaemia and Cancer Research
Foundation, Perth, Western Australia and National Institutes of Health Grants
CA 83088.
62
CHAPTER 3
GENE EXPRESSION LEVELS ASSESSED BY OLIGONUCLEOTIDE
MICROARRAY ANALYSIS AND QUANTITATIVE REAL-TIME
RT-PCR – HOW WELL DO THEY CORRELATE?
3.1 Abstract
Background The use of microarray technology to assess gene expression levels is now
widespread in biology. The validation of microarray results using independent mRNA
quantitation techniques remains a desirable element of any microarray experiment. To
facilitate the comparison of microarray expression data between laboratories it is
essential that validation methodologies be critically examined. We have assessed the
correlation between expression scores obtained for 48 human genes using
oligonucleotide microarrays and the expression levels for the same genes measured by
quantitative real-time RT-PCR (qRT-PCR).
Results Correlations with qRT-PCR data were obtained using microarray data that
were processed using robust multi-array analysis (RMA) and the MAS 5.0 algorithm.
Our results indicate that when identical transcripts are targeted by the two methods,
correlations between qRT-PCR and microarray data are generally strong (r = 0.89).
However, we observed poor correlations between qRT-PCR and RMA or MAS 5.0
normalised microarray data for 13% or 16% of genes, respectively.
Conclusion These results highlight the complementarity of oligonucleotide microarray
and qRT-PCR technologies for validation of gene expression measurements, while
emphasising the continuing requirement for caution in interpreting gene expression
data.
63
3.2 Background
The use of microarray technology to assess gene expression levels is now widespread
in biology and, particularly in the clinical setting, the applicability of the methodology
is likely to broaden as the technology evolves, data analysis procedures improve, and
costs decline (Jordan et al, 2002; Howbrook et al, 2003; Russo et al, 2003). Two
distinct microarray platforms, cDNA and oligonucleotide, are currently in general use
(Kees et al, 2004). While the relative merits of the two systems continue to be
discussed (Moreau et al, 2003), the validation of microarray results using independent
mRNA quantitation techniques, including Northern blotting, ribonuclease protection,
in situ hybridization, or quantitative real-time reverse transcription-polymerase chain
reaction (qRT-PCR) remains a critical element of any microarray experiment
(Brazma et al, 2001; Chuaqui et al, 2002). Despite this, there have been few systematic
validation studies of cDNA, or more noticeably, oligonucleotide microarray data using
these independent approaches. For researchers to be confident with the interpretation of
microarray results and for the establishment of consistent validation procedures in the
microarray community for the purpose of data comparison, it is important that this
issue be addressed.
We have undertaken an extensive series of experiments examining gene expression
profiles in paediatric cancer specimens and normal tissues using oligonucleotide
microarrays. For these studies, we used HG-U133A GeneChips (Affymetrix) which
contain 22,283 probe sets representing approximately 14,500 human genes. To
determine the preferred methodology for the analysis of our microarray data we
compared the correlation between microarray expression scores obtained using two
different data normalisation procedures – Affymetrix MAS 5.0 (Affymetrix technical
note #1), and robust multi-array analysis (RMA) (Irizarry et al, 2003a) – with the
64
expression levels obtained from follow-up verification experiments using qRT-PCR
(Livak et al, 1995; Heid et al, 1996; Mocellin et al, 2003).
We found that the correlation between qRT-PCR and microarray expression data is
generally strong. While our results highlight the complementarity of oligonucleotide
microarray and qRT-PCR technologies for validation of gene expression
measurements, the poor correlations that we observed for 13–16% of genes emphasises
the importance and continuing requirement for caution in interpreting gene expression
data.
3.3 Results
We have assessed the degree of correlation between microarray expression scores
obtained for 48 genes using HG-U133A GeneChips with expression levels measured
for the same genes using qRT-PCR. The genes that we assessed were identified as part
of a larger study underway in the laboratory examining differential gene expression in
paediatric leukaemias and brain tumour specimens. The 48 genes were targeted for
validation either on the basis of their differential expression between our subsets of
interest (e.g. brain tumour vs normal brain specimens, leukaemia specimens vs normal
CD34+ stem cells) as determined by microarray analysis, or because they mapped to
chromosomal regions of interest. In those cases where there were multiple microarray
probe sets for particular genes, only data from those that showed evidence of
differential expression were chosen for validation. For genes that were selected from
chromosomal regions of interest and not necessarily on the basis of differential
expression, correlations were carried out using data from the probe set deemed most
specific for the gene of interest by the Affymetrix software (e.g. microarray probe sets
designated -at are considered more specific than -s-at and -x-at probe sets).
65
In total, 889 specimen/gene combinations were assayed by qRT-PCR and microarray in
this study. Overall, statistically significant correlations (p < 0.05) were observed
between qRT-PCR and RMA normalised data for 33/48 (69%) genes, and between
qRT-PCR and MAS 5.0 normalised data for 32/48 (67%) genes (Tables 3.1 and 3.2,
genes in bold). Typical data for a gene with a good correlation is presented in Figure
3.1. The correlation between the qRT-PCR data and microarray data normalised using
either of the two methods was not significant (p > 0.05) for 14/48 (29%) genes
(Tables 3.1 and 3.2, genes non-bold). Two genes, FLJ20003 and RB, showed
significant correlations by RMA but not by MAS 5.0 analysis, while one gene, GCLC,
had a significant correlation by MAS 5.0 but not by RMA.
By careful analysis of the relevant databases (see Methods) we identified a subset of 31
genes for which the microarray probe-sets were deemed to recognise the exact same
transcript or subset of transcripts as the qRT-PCR probes (Table 3.1). When we
assessed the levels of correlation for this group of 31 transcript-concordant genes a
higher proportion of significantly correlating scores was observed; 84% (26/31) for
MAS 5.0 normalised data and 87% (27/31) for RMA normalised data (Table 3.1, genes
in bold). In addition, the average correlations between the MAS 5.0 or RMA data and
the qRT-PCR data for this subset of genes were very similar (0.71 and 0.72,
respectively). In contrast, for the remaining 17 genes for which the Affymetrix
microarray probe-sets may not recognise the same subset of transcript(s) recognised by
qRT-PCR probes, significant correlations were observed for only 41% (7/17) genes by
either MAS 5.0 and RMA (Table 3.2). All genes with poor correlations were tested on
the same numbers of samples as those genes that did correlate, and there was no
relationship between sample type and whether or not correlation was significant.
Separate genes were targeted for each sample type. Using a two sample t-test, the
average correlations between RMA-qRT-PCR scores and MAS-qRT-PCR scores for
66
the transcript concordant genes in Table 3.1 were significantly higher than the average
of the equivalent correlations for the non-concordant genes in Table 3.2
(RMA-qRT-PCR Table 3.1 vs 3.2, p = 0.0005; MAS-qRT-PCR Table 3.1 vs 3.2,
p = 0.0003).
Determining fold-changes in gene expression levels between subsets of interest is often
a major aim of microarray studies. To address this issue, we analysed fold-change in
average gene expression levels between our subsets of interest (e. g. tumour vs normal)
by both qRT-PCR and RMA or MAS 5.0 microarray scores for the same genes. Only
the 31 transcript-concordant genes were considered in this analysis (Table 3.1). From a
total of 587 specimen/gene combinations we found a significant and strong correlation
in mean fold-change using both RMA (r = 0.89, p < 0.05) and MAS 5.0 (r = 0.92,
p < 0.05) (Figure 3.2a, b). Interestingly, we noticed a trend towards poorer correlation
for genes that exhibited fold-change differences of <1.5 between subsets of interest
based on microarray expression scores compared to those with fold-change differences
of >1.5 (data not shown). The slopes of the two regression lines in Fig. 3.2 are
significantly greater than one [RMA vs qRT-PCR = 1.49 (95%CI = 1.20, 1.77);
MAS vs qRT-PCR = 1.23 (95% CI = 1.03, 1.42)].
67
Table 3.1 A comparison of average qRT-PCR, RMA, and MAS 5.0 scores and the
corresponding correlation values for the 31 transcript-concordant genes assayed
in this study for which the Affymetrix microarray probesets (Affy IDs) were
deemed likely to recognise identical transcripts to qRT-PCR probes. Genes are
ranked from lowest to highest average log2 RMA scores. Genes with significant
correlations (p < 0.05) obtained by either normalisation procedure are highlighted in
bold. The number of specimens tested for each gene is included (n). Expression levels
are shown as log2.
GENE EXPRESSION CORRELATION
NAME AFFY ID n RMA MAS 5.0
qRT-PCR
RMA-qRT-PCR
MAS-qRT-PCR
LCE 204256_at 22 4.79 7.01 0.27 0.81 0.70 ALDH1A1 212224_at 22 4.93 6.66 -2.93 0.89 0.88
CFLAR 211317_s_at 13 5.61 7.92 -0.12 0.65 0.75 REL 206036_s_at 13 5.63 8.36 0.53 0.76 0.77
ABCC4 203196_at 22 5.84 7.54 0.09 0.78 0.89 FOXO1A 202724_s_at 19 5.91 7.43 -2.05 0.85 0.90 NOTCH2 212377_s_at 13 6.19 8.24 -0.43 0.77 0.82
TNFRSF21 214581_x_at 13 6.22 8.03 1.98 0.83 0.97 MADH9 206320_s_at 19 6.24 5.47 -2.29 0.87 0.74 PPM1D 204566_at 30 6.29 8.80 0.50 0.73 0.72 MAP7 202889_x_at 22 6.42 6.27 -3.51 0.85 0.87
DMBT1 208250_s_at 19 6.49 7.25 -6.49 0.20 -0.11 SNIP1 219409_at 13 6.57 8.34 1.08 0.69 0.77 OSF2 210809_s_at 19 6.59 7.68 -1.26 0.80 0.77
ATBF1 208033_s_at 19 6.64 7.14 0.27 0.81 0.84 KIT 205051_s_at 22 6.70 7.51 -2.73 0.86 0.87 P53 201746_at 19 7.00 8.50 -3.44 0.41 0.11
BAG3 217911_s_at 19 7.04 8.61 -0.98 0.79 0.82 RB 203132_at 19 7.04 9.14 -2.82 0.45 0.38
WBP4 203599_s_at 19 7.28 8.97 -0.24 0.62 0.74 BNIP2 209308_s_at 13 7.58 9.79 0.56 0.68 0.69
UMPCMPK 217870_s_at 13 8.17 10.98 1.10 0.37 0.12 DCAMKL1 205399_at 19 8.18 9.23 -3.57 0.76 0.89
OAZIN 201772_at 30 8.22 10.36 -0.36 0.72 0.77 LHFP 218656_s_at 19 8.37 9.27 -0.46 0.89 0.90 BTG3 205548_s_at 13 8.47 10.54 0.83 0.86 0.90 DCX 204850_s_at 19 8.81 10.08 0.62 0.87 0.88
TERF2 203611_at 19 9.05 10.04 -0.14 0.32 0.31 GADD45A 203725_at 19 9.17 9.80 -0.12 0.96 0.94
PRSS11 201185_at 19 9.22 9.85 -3.54 0.63 0.64 RAP1 201174_s_at 19 10.34 11.59 -0.82 0.83 0.84
68
Table 3.2 A comparison of average qRT-PCR, RMA, and MAS 5.0 scores and the
corresponding correlation values for the 17 genes assayed in this study for which
the Affymetrix microarray probesets (Affy IDs) may not recognise the exact same
transcript subsets recognised by qRT-PCR probes. Genes are ranked from lowest to
highest average log2 RMA scores. Genes with significant correlations (p < 0.05)
obtained by either normalisation procedure are highlighted in bold. The number of
specimens tested for each gene is included (n). Expression levels are shown as log2.
GENE EXPRESSION CORRELATION
NAME AFFY ID n RMA MAS 5.0
qRT-PCR
RMA-qRT-PCR
MAS-qRT-PCR
CDC14A 210742_at 13 5.77 7.64 -0.67 0.31 0.26 P125 209175_at 19 6.61 8.43 0.54 0.11 -0.11
GCLC 202922_at 13 6.65 9.20 0.33 0.46 0.56 MAP3K7 206853_s_at 13 6.65 8.72 1.19 0.11 -0.10
TIAL1 202405_at 19 6.68 7.86 0.30 0.32 0.17 FLJ20003 219067_s_at 19 6.71 8.54 0.64 0.64 0.34
RUNX1 210365_at 13 6.95 9.24 1.47 0.29 0.28 PLEKHA1 219024_at 19 6.99 8.18 -2.88 -0.40 -0.28 FLJ12661 218420_s_at 19 7.35 8.50 0.57 -0.08 -0.17
RGC32 218723_s_at 19 7.36 8.11 -3.20 0.85 0.96 WDR11 218090_s_at 19 7.96 9.04 0.60 0.12 0.01 RFC3 204127_at 19 8.10 9.77 1.20 0.62 0.64 ASAH1 213702_x_at 22 8.30 10.29 1.46 0.29 0.27 P38IP 220408_x_at 19 8.35 9.55 0.76 0.73 0.65 BUB3 201456_s_at 19 8.41 9.70 0.60 0.64 0.61 SAC2 203607_at 19 8.86 10.21 0.35 0.22 0.12
TSC22 215111_s_at 19 10.56 11.87 -1.34 0.83 0.82
69
Figure 3.1 Examples of Pearson's correlations between gene expression levels
determined by qRT-PCR and oligonucleotide microarray for one gene assessed in
this study. The mRNA levels for the gene GADD45A were determined by qRT-PCR
and correlated with microarray expression scores determined after data processing
using MAS 5.0 software (A) or RMA (B). All data are shown as log2.
70
Figure 3.2 Pearson's correlations between fold-change in average gene expression
levels between subsets of interest assessed by qRT-PCR and either MAS 5.0
software (A) or RMA (B) for the 31 transcript-concordant genes (see Table 3.1).
All data are shown as log2.
71
3.4 Discussion
Microarray expression analysis has revolutionised many facets of biology and will
continue to be applied widely. However, significant questions remain with regard to the
generation, analysis, and in particular, interpretation of microarray data. Although the
validation of microarray expression results obtained for specific genes using
independent techniques is still considered a desirable component of any microarray
experiment, the genes selected for validation a priori, are usually identified from the
microarray data. The selection is based on the implicit assumption that there is a good
correlation between the microarray data and actual mRNA levels in the cells under
investigation. One fundamental issue that has not been adequately addressed is how
well microarray expression scores reflect actual mRNA levels in the sample being
examined.
To facilitate data comparison between research groups it is important that the
microarray community moves to adopt consistent validation methodologies. This is
especially important if microarray technology is to play a role in the clinical setting
(Petricoin et al, 2002). However, the choice of validation methodology remains a
contentious issue (Rockett et al, 2004). To date, qRT-PCR is the method of validation
that has been used in the majority of published microarray studies, presumably because
it is a rapid, sensitive, high throughput procedure that requires minimal amounts of test
material compared to techniques such as Northern blotting or ribonuclease protection
assays. As is the case for many studies, including ours, qRT-PCR is often the only
feasible approach when rare or unique tissues are investigated. For these reasons, it
would appear likely that qRT-PCR will continue to be used extensively for the
validation of microarray expression data (Klein et al, 2002). To our knowledge, this
study is the most extensive and practical examination of mammalian cells that focuses
72
on the degree of correlation between expression level measurements obtained by
oligonucleotide microarray analysis and qRT-PCR.
We observed strong correlations (p < 0.05) for the majority (>87%) of the 31
transcript-concordant genes that we examined in this study. In addition, although the
MAS 5.0 software and RMA use different algorithms for the normalisation of
microarray data (Affymetrix technical note #1; Irizarry et al, 2003a) we found that the
degree of correlation between microarray and qRT-PCR results was very similar
irrespective of the normalisation procedure employed.
Our data clearly demonstrate that similar microarray scores for different genes do not
necessarily mean that similar qRT-PCR scores will be obtained. For example, ATBF1,
OSF2, and SNIP1 yielded similar average log2 RMA scores (~6.6) but the average log2
qRT-PCR scores for the same genes were substantially different (0.27, -1.26, and 1.08,
respectively). Similarly, KIT and ABCC4 exhibited identical average log2 MAS 5.0
scores (~7.5), while the corresponding average log2 qRT-PCR scores were -2.73 and
0.09, respectively. The finding that genes with similar microarray expression scores
were unlikely to have similar qRT-PCR results presumably reflects the different
hybridisation kinetics of the probe sets for each gene. This observation has the major
implication that on the basis of the qRT-PCR data that we obtained, it is generally not
feasible to predict the true expression level of one gene based on the microarray
expression score of another. In addition, we observed significant correlations for many
genes with microarray expression scores, at least by RMA, of less than 100
(~log2100 = 6.64), which is at the lower end of the range of microarray scores we
obtained in this study (range 6–23000). This finding indicates that the exclusion of
genes with low microarray expression scores (e.g. <100) from further analysis, as has
been adopted by some research groups in early microarray studies, may not be justified.
73
Determining fold-changes in gene expression levels between subsets of interest is often
a critical aim of microarray studies. We found a significant and strong correlation using
RMA (r = 0.89, p < 0.05) and MAS 5.0 (r = 0.92, p < 0.05). These data indicate that the
direction of change of gene expression levels (i.e. either up or down regulation)
between subsets of interest is accurately predicted by comparison of average
microarray expression scores. Again, the fold-change correlations we observed were
very similar irrespective of the normalisation procedure we employed. Consistent with
the results of Yuen et al (2002), fold change results determined by qRT-PCR were
significantly greater than fold change assessed for the same genes by microarray
analysis.
A recent study addressing gene expression profiles in Arabidopsis reported a good
correlation between oligonucleotide microarray and SYBR green qRT-PCR data when
ratios of gene expression in shoot tissue versus root tissue were compared for highly
expressed genes. However, the correlations between shoot versus root ratios were
generally poor for genes expressed at low levels (Czechowski et al, 2004). We
observed a similar trend towards poorer correlation for genes that exhibited fold-change
differences of <1.5 between subsets of interest based on microarray expression scores
compared to those with fold-change differences of >1.5. It is likely that this trend
relates to the fact that small variations in mRNA levels (<2-fold) can be accurately
detected by qRT-PCR, while the smaller dynamic range of microarrays means that the
same changes may not be accurately reflected by microarray expression scores,
especially for genes expressed at low levels (<1.5 pM or approximately 3.5 copies/cell)
(Affymetrix technical note #2; Mutch et al, 2002). This latter point is a likely
explanation for the poor correlation observed for one gene, DMBT1, which is expressed
at very low levels according to our qRT-PCR data. Etienne et al., (2004) observed a
lower overall correlation between microarrray and semi-quantitative RT-PCR data
74
compared to our study. These authors hypothesised that in addition to genes with low
expression levels, those with very high expression levels or a greater percentage of
absent calls, may show lower levels of correlation between Affymetrix expression
scores and semi-quantitative RT-PCR data. We considered these issues in relation to
the other poorly correlating genes in our study and found that none were expressed at
levels that approach the fluorescence ceiling for the Affymetrix scanner (~50000). In
addition, the absolute number or percentage of absent calls did not correlate
significantly (p > 0.05) with the level of correlation between qRT-PCR results and
microarray data (data not shown). It is possible that the differences between our results
and those of Etienne and co-workers are related to the particular semi-quantitative
RT-PCR methodology employed by these researchers, which may not be as sensitive as
qRT-PCR, and as the authors point out, may not detect certain low level transcripts.
In addition to DMBT1 mentioned above, we identified 13 other poorly correlating
genes from the 48 genes we assessed. Careful analysis of the alternative transcript data
available through the LocusLink database http://www.ncbi.nih.nlm/LocusLink
indicated that for 10 of these 13 genes, different subsets of alternative transcripts may
be recognised by microarray probe sets and qRT-PCR probes. Hence, this may be the
explanation for the poor correlations observed for these genes. Possible explanations
for the poor correlations that were observed for the three remaining genes (p53,
UMPCMPK, and TERF2), all of which were transcript-concordant, include the
existence of alternative cross-hybridising transcripts differentially recognised by the
oligonucleotide probe sets and qRT-PCR probes, gene specific variation related to the
different hybridisation kinetics associated with the two technologies, and misleading
results associated with errors in GenBank sequence data and/or probe set annotations
(Gilbertson et al, 2003). Additional experimental data will be required to address these
possibilities. It is important to note that in our hands the reproducibility of both the
75
qRT-PCR and oligonucleotide microarray methods is very high (Kees et al, 2001;
Hoffmann et al, 2005). Hence, it is unlikely that poor correlations observed in our study
are associated with issues of experimental precision.
Interestingly, the microarray and qRT-PCR expression data correlated well for five
genes for which the microarray probe sets were deemed unlikely to recognise the same
transcripts as the qRT-PCR probes. These data suggest that despite the possibility of
differential transcript recognition, identical transcripts were being detected by both
assays in the particular tissues involved.
3.5 Conclusion
Our data indicate that correlations between qRT-PCR and microarray data are generally
strong; a result that is particularly encouraging for those researchers with access to only
very limited amounts of rare or unique test specimens. Our data also emphasise the
importance of ensuring that qRT-PCR probes recognise the same transcript(s) as the
microarray probe set. Finally, the 13–16% non-concordance that we observed indicates
that independent validation of expression data continues to be an important
consideration.
3.6 Methods
3.6.1 Specimens
Informed consent for the use of tissues for research purposes was obtained for all
individuals involved in this study according to hospital and Australian National Health
and Medical Research Council (NHMRC) guidelines.
76
We extracted total RNA from 64 specimens, including 13 primary paediatric brain
tumours, six paediatric brain tumuor cell lines, two normal adult brain cortices, and one
foetal brain germinal matrix. We also obtained total RNA from foetal brain pooled
from multiple individuals (Clontech). In addition, total RNA was extracted from 36
paediatric acute lymphoblastic leukaemia bone marrow specimens and from CD34+
haematopoietic stem cells isolated from the bone marrows of 5 normal individuals.
Ficoll-hypaque purified leukaemia cells or cryopreserved bone marrow specimens were
snap frozen and stored in liquid nitrogen until required. Total RNA was extracted from
~1 × 106 – 2 × 107 live cells. Primary brain tumour specimens (10 – 150 mg) were
either wrapped in foil or placed in RNAlater (Ambion) immediately after resection and
stored at -80°C. Brain tumour cell lines were processed directly from tissue culture.
3.6.2 RNA extraction, preparation of target cRNA and hybridisation to
HG-U133A GeneChips
Total RNA was extracted from all specimens using a combination of TRIZOL reagent
(Invitrogen), RNeasy Mini kit (Qiagen) and ethanol precipitation. Following the
TRIZOL reagent procedure, 0.53 volumes of 100% ethanol were added drop-wise to
the aqueous phase and the mixture applied to RNeasy mini columns according to the
manufacturer's instructions. Further purification and concentration was achieved
through an additional ethanol precipitation. The integrity of the RNA preparation was
assessed using agarose gel electrophoresis and analysis on an Agilent 2100 Bioanalyser
(Agilent Technologies). Biotinylated cRNAs for hybridisation were prepared from total
RNA according to Affymetrix protocols. Agarose gel electrophoresis was used to
confirm the integrity of labelled cRNA and to assess its fragmentation products.
Biotinylated cRNA preparations (15 μg) were hybridised to HG-U133A arrays, which
77
were subsequently washed, stained, and scanned using a GeneArray Scanner (Agilent
Technologies) according to the Affymetrix protocol.
3.6.3 Processing and statistical analysis of microarray data
Array images were reduced to intensity values for each probe (cel files) using
Affymetrix MAS 5.0 software and only those microarrays meeting acceptable
Affymetrix quality control criteria were considered for further analysis. Cel files were
then processed using either the MAS 5.0 software (Affymetrix technical note #1) or
RMA (Bioconductor release 1.2) (Irizarry et al, 2003a), an alternative algorithm that is
publicly available at http://www.bioconductor.org. The MAS 5.0 algorithm uses a
scalar normalisation technique taking into account perfect match (PM) and mismatch
(MM) probe pairs to correct for non-specific hybridisation, while RMA is based on a
quantile normalisation approach which ignores MM values. All microarrays processed
using the MAS 5.0 software were scaled to a standard target intensity of 500. For
comparison purposes, all microarray and qRT-PCR data are presented as log2 and
absent/present calls generated by the MAS 5.0 software were not taken into account.
Pearson's correlations were used for the comparison of qRT-PCR and microarray data
and p-values were obtained using Fisher's z-transformation. Correlations were
considered significant at p < 0.05.
3.6.4 Bioinformatics
To determine whether transcripts recognised by microarray probe sets (Liu et al, 2003)
were likely to be identical to those detected by qRT-PCR probes, alternative splicing
patterns for each gene were thoroughly reviewed using LocusLink
http://www.ncbi.nlm.nih.gov and Ensembl http://www.ensembl.org. Any full-length
78
human mRNA or cDNA sequences demonstrating alternative splicing, in addition to
NCBI-reviewed Reference Sequences (RefSeq), were considered as potential isoforms
for each gene. Using BLAST alignments http://www.ncbi.nlm.nih.gov of probe and
cDNA sequences, the members of each isoform 'family' that could be targeted by either
qRT-PCR or microarray were identified (typically multiple isoforms for each gene).
The potential number of isoforms recognised by each technology were then compared.
Probes which targeted exactly the same isoform subsets for each gene were considered
'transcript-concordant' and placed in Table 3.1; those for which at least one of the
targeted isoforms differed (regardless of the number of matching isoforms) were
considered 'non transcript-concordant' and placed into Table 3.2.
3.6.5 qRT-PCR
All qRT-PCR assays were carried out using primer and probe sets from Applied
Biosystems (ABI Assays on Demand, http://www.appliedbiosystems.com/). Each assay
was designed using ABI's primer/probe selection algorithm and bionformatics pipeline
which includes access to both public and Celera DNA sequence databases. The
combination of gene specific primers and a gene specific probe ensures a high degree
of specificity.
Aliquots of total RNA extracted for microarray analysis as described above were used
for qRT-PCR experiments according to the manufacturer's protocols (ABI). All ABI
Assays on Demand are designed to generate amplicons of 50–150 bp and are carried
out using identical cycling conditions. 1–2 ug total RNA (quantitated by
spectrophotometer at OD260) was used for each RT reaction. Three RT reactions were
pooled and all qRT-PCR reactions were carried out using aliquots from the pool. We
did not detect DNA contamination in any of our total RNA preparations after
79
qualitative assessment using an Agilent Bioanalyzer. All qRT-PCR assays for a
particular gene were undertaken at the same time under identical conditions and carried
out in duplicate. All qRT-PCR experiments were run on an ABI 7700 sequence
detector.
For all qRT-PCR assays the expression levels of target genes were normalised to the
levels of the ACTB housekeeping gene utilising a standard curve method for
quantitation as described previously (Kees et al, 2003a). Serial dilutions of cDNAs
generated from selected cell lines that expressed target genes at a suitable level were
used to generate a standard curve for each target gene and ACTB. The standard curves
were then used to determine expression values (expressed as ng cDNA template) for
each target gene after qRT-PCR analysis of each test specimen. Relative expression
values for each target gene were expressed as a ratio of target gene expression level to
ACTB expression level in the same specimen. These ratios were then correlated with
the microarray data.
3.7 Author contributions
PBD and NGG contributed equally to this work and were responsible for designing the
study, analysing, collating, and interpreting the data, and preparing the manuscript.
MJF carried out the statistical analysis, AHB and KF assisted with data analysis,
experimental design, and data interpretation. PAT, JRF, JMB, AJC and NGG carried
out the microarray and qRT-PCR experiments. URK supervised all aspects of the study
and preparation of the manuscript.
80
3.8 Acknowledgements
This study was supported by funds from NHMRC project grants 254595 and 254596,
NCI/NIH grant 95475, the Three Boys Legacy, and Variety Club of Western Australia.
We would like to thank Nigel Swanson and Violet Peeva at the Lotterywest State
MicroArray Facility, Perth, Western Australia. Thanks also to Reinete Orr for
secretarial assistance. NGG was supported by a National Childhood Cancer Foundation
Laura and Greg Norman Fellowship.
81
CHAPTER 4
IDENTIFICATION OF NOVEL PROGNOSTIC MARKERS FOR
PAEDIATRIC T-CELL ACUTE LYMPHOBLASTIC LEUKAEMIA
4.1 Abstract
In the last four decades the survival of patients with newly diagnosed childhood T-cell
acute lymphoblastic leukaemia (T-ALL) has improved dramatically. In sharp contrast,
relapsed T-ALL continues to confer a dismal prognosis. We sought to determine if
gene expression profiling could uncover a signature of outcome for children with
T-ALL. Using 12 patient specimens obtained before therapy started, we examined the
gene expression profile by oligonucleotide microarrays. We identified three genes,
CFLAR, NOTCH2 and BTG3, whose expression at the time of diagnosis accurately
distinguished the patients according to disease outcome. These genes are involved in
the regulation of apoptosis and cellular proliferation. The prognostic value of the three
predictive genes was assessed in an independent cohort of 25 paediatric T-ALL patients
using quantitative real-time reverse transcription polymerase chain reaction. Patients
assigned to the adverse outcome group had a significantly higher cumulative incidence
of relapse compared with patients assigned to the favourable outcome group
(46% vs. 8%, P = 0.029). Five-year overall survival was also significantly worse in the
patients assigned to the adverse outcome group (P = 0.0039). The independent
influence of the 3-gene predictor was confirmed by multivariate analysis. Our study
provides proof of principle that genome-wide expression profiling can detect novel
molecular prognostic markers in paediatric T-ALL.
82
4.2 Introduction
In the past four decades dramatic advances have been made in the treatment of
childhood T-cell acute lymphoblastic leukaemia (T-ALL), with cure rates of
approximately 75% now being achieved (Goldberg et al, 2003). However, despite the
use of intensive risk-adapted chemotherapy, treatment failure occurs in approximately
25% of patients, and outcome for this group remains dismal (Goldberg et al, 2003;
Einsiedel et al, 2005). A major goal therefore was to identify at diagnosis patients at
high risk for relapse, in order that intensified treatment and/or novel therapeutic
approaches may be offered. Unfortunately, clinical presenting features have proven to
be much less reliable predictors of outcome in T-ALL compared with pre-B ALL
(Pullen et al, 1999). Pre-B ALL is characterised by several chromosomal translocations
that are strongly associated with patient outcome (Heerema et al, 1998), hence are used
to guide treatment intensity. In contrast, no molecular markers are currently applied to
stratify patients with T-ALL. Recently, molecular analysis of T-ALL revealed that the
majority of cases could be subgrouped according to the expression of several
transcription factor oncogenes (Ferrando et al, 2002). These molecular subgroups are
linked to the immunophenotypic features of normal thymocytes arrested at different
stages of differentiation (Ferrando et al, 2002). Importantly, several of the transcription
factors defining these distinct molecular subgroups, including HOX11 (TLX1),
HOX11L2 (TLX3), TAL1 (SCL) and LYL1 appear to have prognostic significance in
T-ALL (Ferrando et al, 2002). The outcome for patients whose T lymphoblasts
overexpress HOX11 is generally considered favourable (Ferrando et al, 2002, 2004b;
Kees et al, 2003a). On the other hand, the prognostic value of HOX11L2 remains
equivocal and may depend upon the therapeutic regimen utilised (Ballerini et al, 2002;
Ferrando et al, 2002; Cavé et al, 2004; Gottardo et al, 2005). A global signature
predictive of outcome for paediatric T-ALL would therefore be of immense clinical
83
use. To uncover novel molecular prognostic markers for childhood T-ALL we
performed gene expression profiling of diagnostic bone marrow specimens and
compared the data from patients who remained in continuous complete remission
(CCR) to those who relapsed.
4.3 Methods
4.3.1 Patient characteristics
Informed consent was obtained for all individuals involved in this study. Paediatric
T-ALL patients were divided into training (n = 12) and validation (n = 25) cohorts
based upon the quantity of bone marrow material available. Specimens with
insufficient material for microarray were allocated to the validation cohort [the latter
tested only by quantitative real-time reverse transcription polymerase chain reaction
(qRT-PCR)]. Specimens were obtained at diagnosis before therapy started and all
patients were treated on Children's Cancer Group (CCG) risk-adjusted protocols
between 1985 and 2002. Characteristics of the training and validation cohorts are
shown in Table 4.1. There were no significant differences in these parameters except
for median age, which was significantly lower in the training cohort
(P = 0·014, Mann–Whitney U-test). However, the age ranges of patients were similar
(Table 4.1).
84
Table 4.1 Patient characteristics. Percentages are shown in parenthesis.
Training Cohort N = 12
Validation cohort N = 25
Follow up (yrs) of patients in CCR
Median 8.2 4.5 Range 5.5 – 10.5 3.2 – 12.8
Sex Male: Female 9 (75): 3 (25) 20 (80): 5 (20)
Age (yrs) Median 6.6 13.1* Range 2.0 – 13.4 1.8 – 17.8
WBCa x109 /L Median 59.5 227 Range 9 – 740 16 – 791
NCIb Standard risk 1 (8) 1 (4) High risk 11 (92) 24 (96)
Cytogenetics Normal karyotype 3 (25) 5 (20) Abnormal karyotype 8 (67) 5 (20) Pseudodiploid 5 (42) 5 (20) Hyperdiploid 3 (25) 0 Missing/Inadequate 1 (8) 15 (60)
Status of day 7 BMc RERd 10 (83) 13 (52) SERe 2 (17) 11 (44) Not performed 0 1 (4)
Induction result M1f 11 (92) 22 (88) M2g 1 (8) 2 (8) M3h 0 0 Not performed 0 1 (4)
Total relapses (%) 5 (42) 7 (28) Time to relapse (yrs)
Median 2.0 1.4 Range 0.45 – 2.55 0.42 – 2.46
aWBC indicates white blood cell count; bNCI, National Cancer Institute; cBM, bone marrow;
dRER, Rapid early responder (<25% blasts in BM at day 7);
eSER, Slow early responder (>25% blasts in BM at day 7);
fM1, <5% blasts in BM; gM2, 5 – 25% blasts in BM; hM3, >25% blasts in BM
*P = 0.0014 by Mann-Whitney U test
85
There were no statistically significant differences between the training and validation
cohorts with regard to outcome. Five-year relapse-free survival (RFS) was 58%
[standard error (SE) ± 14.2%] for the training cohort and 72% (SE ± 8.9%) for the
validation cohort (P = 0.45). Similarly, the 5-year overall survival (OS) was 58%
(SE ± 14.2%) for the training cohort and 65% (SE ± 9.5%) for the validation cohort
(P = 0.59). These outcomes are in line with results achieved over the same time period
for large patient cohorts treated on these protocols (Gaynon et al, 2000; Seibel et al,
2004).
4.3.2 RNA extraction, preparation of target cRNA and hybridisation to
HG-U133A GeneChips
The methods have been described in detail previously (Hoffmann et al, 2005). RNA
was extracted from bone marrow specimens using a modified TRIZOL method
(Invitrogen, Carlsbad, CA, USA) and ethanol precipitation. The integrity of the RNA
preparation was assessed using agarose gel electrophoresis and analysis on an Agilent
2100 Bioanalyser (Agilent Technologies, Palo Alto, CA, USA). Biotinylated cRNAs
for hybridisation were prepared from 2 μg of total starting RNA, following an adapted
Affymetrix protocol (Hoffmann et al, 2005). Biotinylated cRNA preparations (15 μg)
were hybridised to HG-U133A GeneChips (Affymetrix, Santa Clara, CA, USA) in
accordance with Affymetrix protocols. Array data are available through Array Express
(http://www.ebi.ac.uk/arrayexpress).
4.3.3 Quantitative real-time reverse transcription polymerase chain reaction
To verify the expression level of selected genes, qRT-PCR was employed. The detailed
methodology has been previously described (Dallas et al, 2005). Briefly, qRT-PCR
86
assays were carried out using primer and probe sets from Applied Biosystems, Foster
City, CA, USA (ABI Assays on Demand, http://www.appliedbiosystems.com/).
Aliquots of total RNA extracted for microarray analysis as described above were used
for qRT-PCR experiments according to the manufacturer's protocols (Applied
Biosystems). qRT-PCR assays were run in duplicate on an ABI 7700 sequence detector
(Applied Biosystems). The expression levels of target genes were normalised to the
levels of the ACTB housekeeping gene utilising a standard curve method for
quantitation.
4.3.4 Statistical analysis and bioinformatics
Our gene filtering approach and the detection of differential expression utilising a
supervised tree-based algorithm called a Random Forest (RF) has been described in
detail (Beesley et al, 2005; Hoffmann et al, 2006). Briefly, a variance filter was applied
to remove probe sets with an absolute fold change <1·15 and an associated P-value of
>0·1 by permutation test (999 permutations). The RF algorithm was applied to rank
probe sets with respect to their ability to discriminate between adverse and favourable
outcome groups.
To generate an outcome prediction model we developed a classifier algorithm using
principal component analysis (PCA), in a manner similar to Lugthart et al (2005). This
model was developed using qRT-PCR expression data. PCA was applied to the
qRT-PCR values obtained for the best combination of genes as measured in the training
set. A line of segregation was generated to optimally divide adverse and favourable
outcome patients. The equations underlying the principal components (PCs) for the
training set were applied to qRT-PCR expression values measured in the validation
cohort (the first PC was −0·5875 × CFLAR + 0·3702 × BTG3 − 0·7196 × NOTCH2; the
second PC was 0·5586 × CFLAR + 0·8289 × BTG3 − 0·0296 × NOTCH2; where
87
CFLAR, NOTCH2 and BTG3 represent continuous qRT-PCR gene expression scores).
This model was thus used to assign specimens to adverse or favourable outcome
groups. Cumulative incidence of relapse, RFS and OS were estimated by the
Kaplan–Meier method and the log-rank test was used for comparison of differences in
outcome between groups. Multivariate analysis was performed using the Cox
regressional model. Pearson's correlations were used for the comparison of qRT-PCR
and microarray data. Chi-squared test and Fisher's exact test were used to assess patient
characteristics with respect to outcome.
Functional analysis of top RF-ranked probe sets was performed using the NetAffx
Gene Ontology (GO) Mining Tool
(http://www.affymetrix.com/analysis/netaffx/index.affx). For this analysis we focused
on the GO Biological Process category. The chi-squared test was used to assess genes
for overrepresentation (P-value: <0.05) of GO categories compared with probe sets on
the HG-U133A GeneChip.
For in silico analysis, we utilised two publicly available data sets from previously
published studies (Yeoh et al, 2002; Chiaretti et al, 2004). CEL files were downloaded
from http://www.stjuderesearch.org/data/ALL1/index.html (Yeoh et al, 2002) and
http://www.bioconductor.org/docs/papers/2003/Chiaretti (Chiaretti et al, 2004). The
CEL files for these data sets were imported and normalised using Robust Multi-Array
Analysis (RMA) in line with our standard methods (Beesley et al, 2005; Hoffmann
et al, 2006). Since data for these two studies was obtained using the HG_U95Av2
GeneChip, the predecessor to the HG-U133A GeneChip utilised in this study, we used
cross-platform comparison spreadsheets from Affymetrix
(http://www.affymetrix.com/support/technical/comparison_spreadsheets.affx) to select
equivalent probe sets between platforms. Where available, probe sets denoted as
‘best-match’ (depicted in bold) or ‘good match’ (underlined) were selected. For our
88
3-gene predictor the following probe sets on the HG_U95Av2 GeneChip were selected
for analysis: 38083_at (NOTCH2), 37218_at (BTG3) and 32746_at (CFLAR). Yeoh
et al (2002) identified seven predictive genes, for which the following probe sets on the
HG-U133A GeneChip were selected for analysis: 208406_s_at (GRAP2), 215043_s_at
(SMA5), 203537_at (PRPSAP2), 212789_at (KIAA0056), 208130_s_at (TBXAS1),
201132_at (HNRPH2) and 212014_x_at (CD44). Chiaretti et al (2004) identified three
predictive genes, for which the following probe sets on the HG-U133A GeneChip were
selected for analysis: 204822_at (TTK), 211986_at (AHNAK) and 205831_at (CD2).
To provide unbiased estimates of the predictive power of the Yeoh et al (2002) and
Chiaretti et al (2004) prognostic genes in our training data set and conversely our
3-gene predictor on their data sets, we utilised the RF to generate a classification
algorithm as previously described (Hoffmann et al, 2006).
4.4 Results
4.4.1 Gene expression profiles of newly diagnosed T-ALL patients
Unsupervised hierarchical cluster analysis using all 22 283 probe sets for the 12 T-ALL
training cohort specimens, clustered patients into two major groups; relapse and CCR,
herein referred to as adverse and favourable outcome, respectively. One specimen
derived from a patient with favourable outcome was found to be separate from the two
groups. Chi-squared testing revealed no significant correlations between outcome and
the clinical parameters age, presenting white blood cell (WBC) count, gender or day 7
bone marrow assessments. To identify genes associated with disease outcome, a RF
algorithm was applied to rank the 3290 probe sets remaining after variance filtering,
with respect to their ability to discriminate between specimens from adverse and
favourable outcome patients. The output list generated by the RF, ranked the probe sets
89
from the most, to the least discriminatory, between adverse and favourable outcome
patients. The RF output list was charted as an importance graph, which plotted
individual probe sets against their discriminatory ability. Probe sets with the greatest
ability occupied the steepest portion of the curve. For further analysis, probe sets were
selected from this portion of the curve, which comprised approximately 300 probe sets.
Using unsupervised hierarchical clustering, the 300 top-ranked probe sets revealed two
distinct clusters according to disease outcome (Fig 4.1). Of the 300 top-ranked probe
sets, 208 (69%) were perfect separators, defined as a gene whose expression level alone
is sufficient to segregate the patients into favourable and adverse outcome groups. The
probability of observing 208 perfect separator probes in the entire data set is
P < 0·00001 (with only 56 perfect separators expected by chance alone), underscoring
the existence of striking intrinsic biological differences present at diagnosis between
T-ALL blasts derived from patients with adverse and favourable outcomes. Of the 300
top-ranked probe sets, 61% were upregulated in cells from patients with an adverse
outcome.
90
Figure 4.1 Unsupervised hierarchical clustering of the training T-ALL cohort
using the 300 top Random Forest-ranked probe sets. Box indicates the favourable
outcome patient that was separate from the two main clusters using all 22 283 probe
sets. WBC, white blood cell count; CCR, continuous complete remission; T-ALL,
T-cell acute lymphoblastic leukaemia.
4.4.2 Functional analysis
To uncover biologically relevant pathways, functional analysis of the 300 top-ranked
probe sets was performed. These 300 probe sets represented 264 individual genes, of
which 189 had annotated biological functions (as of November 2006). Nine GO
categories were found to be significantly overrepresented compared with all probe sets
on the HG-U133A GeneChip (Fig 4.2). Overrepresented categories included genes
involved in cell survival and proliferation (negative regulation of progression through
cell cycle and positive regulation of the I-κB kinase/NF-κB cascade), protein handling
(protein folding, intra-Golgi vesicle-mediated transport and ubiquitin cycle), categories
91
related to gene expression (nucleosome assembly, negative regulation of transcription
from RNA polymerase II promoter and mRNA processing) and metabolism (negative
regulation of metabolism).
The most enriched categories were intra-Golgi vesicle-mediated transport
(P < 0.00001) and positive regulation of the I-κB kinase/NF-κB cascade
(P = 0.000045). The majority of genes (five of seven, 71%) from the positive regulation
of the I-κB kinase/NF-κB cascade GO category, were upregulated in patients with an
adverse outcome and, importantly, remained significantly overrepresented
(P = 0.00006) when only upregulated genes in the adverse outcome group were
assessed for enrichment.
Figure 4.2 Percentage representation according to Gene Ontology biological
function category of 300 top-ranked probe sets (grey bars) compared with the
representation of probe sets on the HG-U133A GeneChip (black bars) using
chi-squared test (P < 0.05 for all nine categories). Reg, regulation; −ve, negative;
+ve, positive
92
4.4.3 Defining a gene expression signature predictive of outcome
We next assessed whether a small set of signature genes, from the 300 top-ranked
probe sets, could predict outcome. Nine genes (Table 4.2) were selected for further
investigation on the basis of statistical criteria (P-value: <0.005 by permutation test,
fold change ≥1.5, in either direction, between adverse and favourable outcome and/or
perfect separator status) in combination with biological function (member of GO
biological process categories identified as overrepresented and/or review of reported
gene functions). None of the nine selected genes have previously been associated with
outcome in paediatric T-ALL. Unsupervised hierarchical clustering using this set of
nine genes revealed a clear discrimination between patients with adverse and
favourable outcomes (data not shown).
Gene symbols depicted in bold had strong and significant correlations
(mean r = 0.78 P ≤ 0.01 by Pearson’s correlation) between expression values measured
by qRT-PCR and microarray. To verify the patterns of gene expression obtained from
microarray analysis, qRT-PCR was employed to measure the expression levels of the
nine selected genes. A strong correlation [mean r = 0.78 (range: 0.65–0.90), P ≤ 0.01]
was observed for all but two genes. Thus, in agreement with previously published
reports (Yeoh et al, 2002; Chiaretti et al, 2004; Dallas et al, 2005), our results confirm
that, in the majority of cases the patterns of gene expression measured by qRT-PCR
and microarray are very similar.
93
Table 4.2 Nine genes selected from the 300 top-ranked probe sets, which
discriminated between adverse and favourable outcome. Genes were selected by
combining statistical criteria (P-value < 0.005 by permutation test, fold change ≥ 1.5, in
either direction, between adverse and favourable outcome and/or perfect separator
status) with biological function (member of GO biological process categories identified
as overrepresented and/or review of reported gene functions).
Gene symbol Gene title HG-U133A Probe set ID
GO Biological function category ΔA/F
CFLAR*
CASP8 and FADD-like apoptosis regulator 211317_s_at
·Positive regn of I-kappaB/ NF-kappaB cascade# ·Induction of/ anti-apoptosis
-1.5
NOTCH2
Notch homolog 2 (Drosophila) 210756_s_at
·Negative regn of progresn
through cell cycle# ·Induction of/ anti-apoptosis
-2.0
BTG3
BTG family, member 3 213134_x_at Negative regn of progresn
through cell cycle# 2.5
SNIP1*
Smad nuclear interacting protein 1 219409_at None ascribed 1.8
RUNX1†* Runt-related transcription factor 1 (acute myeloid leukaemia 1; aml1 oncogene)
208129_x_at Positive regn of transcription 2.8
CDC14A*
CDC14 cell division cycle 14 homolog A (S. cerevisiae) 210743_s_at Regn of progresn through
cell cycle 1.5
REL*
v-rel reticuloendotheliosis viral oncogene homolog (avian) 206036_s_at Positive regn of I-kappaB/
NF-kappaB cascade# 2.1
TNFRSF21*
Tumour necrosis factor receptor superfamily, member 21 214581_x_at ·Apoptosis
·Signal transduction 2.7
BNIP2
BCL2/adenovirus E1B 19kDa interacting protein 2 209308_s_at Anti-apoptosis 1.6
RF indicates Random Forest; GO, Gene Ontology; regn, regulation; progresn, progression;
ΔA/F, fold change in expression between adverse outcome (A) and favourable outcome (F)
*Perfect Separator between adverse and favourable outcome patients
# Significantly overrepresented GO category (P < 0.05 by Chi squared test)
† Gene ranked number 1 by RF
94
To select the optimal combination of genes for predicting outcome from the seven
genes with strong correlations, the qRT-PCR gene expression data were analysed in the
training cohort using PCA (see Statistical analysis and bioinformatics in methods). The
best discrimination of patients according to outcome was achieved using a model based
upon the expression levels of three genes, CFLAR (FLIP) (CASP-8 and FADD-like
apoptosis regulator), NOTCH2 [Notch homolog 2 (Drosophila)] and BTG3 (ANA)
(BTG family, member 3; see Statistical analysis and bioinformatics in Methods). This
model is herein referred to as the 3-gene predictor.
4.4.4 Validation of the signature genes predictive of outcome
The expression level of the three genes was measured by qRT-PCR in a completely
independent validation cohort of 25 paediatric T-ALL patients. For each patient
specimen the measurements were entered into the equation for the 3-gene predictor and
the resulting value used to determine assignment to favourable or adverse outcome
group. Correct identification was recorded for six of the seven patient specimens
obtained from adverse outcome patients and 11 of the 18 favourable outcome patients.
Based on group classification by the 3-gene predictor (favourable or adverse outcome),
patients assigned to the adverse outcome group demonstrated a significantly higher
cumulative incidence of relapse compared with patients assigned to the favourable
outcome group [46% (SE ± 13.8%) vs. 8% (SE ± 7.8%), P = 0.029; Fig 4.3A].
Five-year OS for all patients (training and validation cohorts combined) was
significantly worse for patients predicted with an adverse outcome than for patients
predicted with a favourable outcome [44% (SE ± 11.7%) vs. 82% (SE ± 8.8%),
P = 0.0039; Fig 4.3B]. Univariate analysis revealed no other significant associations
between patient outcomes and other clinical parameters, including age, presenting
WBC count, gender and day 7 marrow examinations. Multivariate analysis adjusting
95
for WBC count <50 × 109 cells/l or >50 × 109 cells/l (or using the cut-off WBC count
<100 × 109 cells/l or >100 × 109 cells/l), age 1 to 9.99 years or ≥10 years and status of
day 7 bone marrow examination [rapid early responder: <25% blasts (M1 or M2) or
slow early responder: >25% blasts (M3)], was conducted. The effect of gender could
not be adequately modelled, as no female patients relapsed. Patients assigned with an
adverse outcome by the 3-gene predictor remained independently associated with an
increased risk of relapse [P = 0.022; hazard ratio (HR) 57; 95% confidence interval
(CI): 1.8–1842] and lower OS (P = 0.0033; HR 12; 95% CI: 2.3–64).
Figure 4.3 A Cumulative incidence of relapse for the validation cohort (n = 25) of
T-cell acute lymphoblastic leukaemia (T-ALL) patients stratified by the 3-gene
predictor [as measured by quantitative real-time reverse transcription polymerase
chain reaction (qRT-PCR)]. B Kaplan–Meier curve of 5 year overall survival
probability for all T-ALL patients (combined cohorts, n = 37) stratified by the 3-gene
predictor (as measured by qRT-PCR). Dashes indicate censored patients.
96
CFLAR and NOTCH2 were generally downregulated (twofold lower level of
expression) in patients with an adverse outcome compared to patients with a favourable
outcome (Fig 4.4A & B). On the other hand, BTG3 was generally upregulated (1.6-fold
higher level of expression) in patients with an adverse outcome (Fig 4.4C).
Figure 4. Mean (SEM) expression levels for genes comprising 3-gene predictor by
quantitative real-time reverse transcription polymerase chain reaction, for
adverse and favourable outcome patients for combined cohorts (n = 37). The
fold-change (ratio of mean expression level for adverse over favourable outcome
groups) is shown
In silico analysis was performed on publicly available data sets previously published by
others, for paediatric (Yeoh et al, 2002) and adult T-ALL patients
(Chiaretti et al, 2004). No statistically significant difference in the expression levels of
our three predictive genes between disease outcome groups was observed in their
T-ALL patients. Further, we conducted the reverse analysis, applying genes identified
in these two studies to our training cohort data set. Genes found to be prognostic by
Yeoh et al (2002) (n = 7) and Chiaretti et al (2004) (n = 3) (see Statistical analysis and
97
bioinformatics in Methods for details of genes and probe sets), were not significantly
differentially expressed between adverse and favourable outcome specimens and were
not predictive of relapse in our training cohort [overall predictive accuracy was 42% for
the Yeoh et al (2002) and 25% for the Chiaretti et al (2004) prognostic genes].
We also examined the prognostic significance of the oncogenic transcription factors
HOX11, HOX11L2, TAL1, LYL1 and OPAL1 (outcome predictor in acute leukaemia 1),
which have previously been associated with disease outcome in paediatric T-ALL
(Ferrando et al, 2002; Mosquera-Caro et al, 2003). In our training cohort the
transcription factor oncogenes and OPAL1 were not predictive of outcome.
4.5 Discussion
Intensified therapeutic regimens, including the use of bone marrow transplantation, fail
to cure the majority of patients with relapsed T-ALL. Identifying patients at high risk of
treatment failure soon after initial diagnosis presents an ideal opportunity to modify
treatment early, either by further treatment intensification and/or the addition of novel
therapies, in an effort to prevent disease recurrence. Currently, it is not possible to
predict which T-ALL patients are most likely to relapse. Although several molecular
markers appear to have prognostic value in childhood T-ALL, none has been
prospectively evaluated and presently no marker is used to stratify patients at diagnosis.
We compared the gene expression profile from a group of children with T-ALL who
remained in CCR (n = 7) with that from a group who relapsed (n = 5). The strong
relationship between disease outcome and gene expression profile supported our
hypothesis that underlying molecular prognostic markers are present at diagnosis.
Consistent with previous reports (Pullen et al, 1999; Goldberg et al, 2003), outcome
was not correlated with other clinical parameters. For future clinical use, to allow
improved patient stratification, it would be advantageous to identify a group of patients
98
at high risk of relapse, by measuring the expression level of a limited panel of genes,
rather than having to conduct more expensive and technically more demanding
microarrays. We selected a set of nine genes, using a combination of statistical and
biological criteria, which accurately grouped patients according to disease outcome
using unsupervised hierarchical clustering. From this set, the qRT-PCR expression
values of three genes, CFLAR, NOTCH2 and BTG3 were found to optimally
discriminate patients according to outcome when combined by PCA modelling.
In a completely independent validation cohort of 25 paediatric T-ALL patients, the
3-gene predictor accurately predicted six of seven adverse outcome patients and 11 of
18 favourable outcome patients. Based on group assignment by the 3-gene predictor
(adverse or favourable outcome) the cumulative incidence of relapse was significantly
higher in the adverse outcome group compared with the favourable outcome group.
Additionally, 5 year OS was significantly worse for patients assigned to the adverse
outcome group. Importantly, multivariate analysis confirmed the 3-gene predictor as an
independent prognostic marker.
Of the eight incorrectly classified patients, only one was a patient who relapsed.
Interestingly, this patient was treated on an earlier era CCG protocol, in the mid-1980s.
In contrast, the remainder of the patients from the validation cohort, as well as the
patients from the training cohort, were treated on more contemporary treatment
protocols from the 1990s onwards. As the therapy utilised to treat a patient is the most
important prognostic factor (Pui & Evans, 2006a), it is possible that the incorrect
classification of this patient could be due to the less intense therapy associated with an
earlier era protocol. In addition, three of the seven incorrectly classified favourable
outcome patients received augmented therapy based on a slow early response to
treatment [defined as >25% blasts (M3) on day 7 bone marrow examination]. One can
speculate that the adverse outcome predicted by the 3-gene predictor may have been
99
averted by early intensification of treatment, an approach demonstrated to improve
outcome in patients with a slow early response to therapy (Nachman et al, 1998).
Our 3-gene predictor was not prognostic in other published data sets using in silico
analysis. As for any diagnostic tests, such tests need to be performed under the same,
quality-controlled conditions, e.g. the same methodology, such as qRT-PCR or
microarrays using the same platform and the exact same experimental protocols
(Sherlock, 2005). In silico testing does not fulfil these requirements and consequently
several factors may explain why our results were not reproducible using in silico
analysis. First, the success may have been hampered by the use of different microarray
platforms [HG_U95Av2 GeneChip in Yeoh et al (2002) and Chiaretti et al (2004)
versus HG-U133A GeneChip in our study] and differences in the signal intensity of
array data of these two studies compared with that in our analysis [the average signal
intensity was 3.54- and 1.69-fold lower than the expression signals on our arrays for the
Yeoh et al (2002) and Chiaretti et al (2004) data, respectively]. Secondly, predictive
genes identified in adult (Chiaretti et al, 2004) and paediatric (Yeoh et al, 2002) T-ALL
studies revealed poor prognostic ability when tested between these analyses,
underscoring the significant biological differences between adult and childhood T-ALL
(Chiaretti et al, 2004). Thirdly, comparing paediatric patients treated on different
therapeutic protocols may also be problematic (Holleman et al, 2006a). Furthermore,
the limited number of patient specimens in the cohorts may have influenced the in
silico analysis.
The gene CFLAR belongs to the positive regulation of the I-κB kinase/NF-κB cascade
GO category, one of the most significantly overrepresented GO categories in our 300
top-ranked probe sets. Constitutive activation of the NF-κB pathway has been reported
as a frequent feature of childhood ALL cells, including T-ALL (11 of 13 specimens),
suggesting this pathway plays a significant role in the survival of leukaemic cells
100
(Kordes et al, 2000). In a large study investigating the role of 70 key apoptosis genes in
childhood ALL, the expression levels of several NF-κB target genes, including CFLAR,
were found to be significantly higher in T-ALL versus pre-B ALL patients
(Holleman et al, 2006b). The authors proposed that NF-κB activity may be upregulated
in T-ALL compared with pre-B ALL (Holleman et al, 2006b). Interestingly, the
pro-apoptotic gene, BCL2L13, the only gene that was independently associated with
patient outcome in the Holleman et al (2006b) study, was also differentially expressed
(P = 0.049) between adverse and favourable outcome patients in our training cohort
although in the opposite direction. CFLAR encodes for an apoptosis regulatory protein
which inhibits the extrinsic apoptotic pathway by blocking the conversion of
procaspase 8 to its active form, caspase 8. However, in certain cellular contexts CFLAR
has been shown to promote apoptosis (Chang et al, 2002; Boatright et al, 2004). Thus,
the precise apoptotic role of CFLAR appears to be related to the cellular context.
CFLAR was generally expressed at lower levels in specimens derived from adverse
outcome patients. Low levels of CFLAR have been associated with an inferior
prognosis in patients with high/intermediate grade non-Hodgkin lymphomas (NHLs)
(Valente et al, 2006) but the opposite has been observed in patients with low-grade
(NHLs) (Valente et al, 2006) and Burkitt lymphoma (Valnet-Rabier et al, 2005).
The other member of the 3-gene predictor, NOTCH2, situated on chromosome
1p13-p11, is a member of the NOTCH type 1 transmembrane receptor protein family.
NOTCH signalling regulates numerous important cellular functions, including cell fate
decisions, differentiation, proliferation and apoptosis (reviewed in Artavanis-Tsakonas
et al, 1999). Notably, aberrant signalling of NOTCH1, NOTCH2 and NOTCH3 has
been implicated in the development of T-ALL/lymphoma (reviewed in Pear & Aster,
2004). Moreover, the recent discovery by Weng et al (2004) that over half of paediatric
patients with T-ALL have activating mutations of the closely related family member,
101
NOTCH1, suggests that NOTCH1 signalling plays a major role in the pathogenesis of
T-ALL. Significantly, the presence of these mutations appears to be associated with
improved patient survival (Breit et al, 2006). We observed generally lower NOTCH2
expression in specimens derived from adverse outcome patients. Low expression of
NOTCH2 has been associated with an inferior prognosis in patients with breast cancer
(Parr et al, 2004).
The third member of the 3-gene predictor, BTG3, belongs to the B-cell translocation
(BTG) antiproliferative protein family (Kawamura-Tsuzuku et al, 2004), which
suppress cell growth by inhibiting the G1 to S transition of the cell cycle (reviewed in
Matsuda et al, 2001). Of note, BTG1, another member of the BTG family, was also
found to be significantly upregulated (P = 0.048) in patients with an adverse outcome.
Our finding is consistent with previous reports that revealed the upregulation of genes
involved in the regulation of cellular proliferation in specimens derived from patients
with good treatment responses (Chiaretti et al, 2004; Cairo et al, 2005).
It is anticipated that in the near future, patient stratification at diagnosis will be
significantly enhanced by the use of molecular profiling, permitting the delivery of
individualised therapy according to the risk of relapse. Despite the relatively small
number of patients investigated in this study and the heterogeneity of treatment
protocols, our data provides proof of principle that genome-wide expression profiling
can identify novel molecular markers for outcome prediction in paediatric T-ALL that
can be validated in an independent cohort. We have also demonstrated that accurate
outcome prediction models can be developed using qRT-PCR expression data,
presently a technique more suitable to high-throughput analysis in a hospital setting.
Our findings warrant further investigation in a larger, uniformly treated cohort of
patients.
102
4.7 Author contributions
NGG and URK conceived the study. The study was performed in URK’s laboratory.
NGG was responsible for analysing, collating, and interpreting the data, carrying out
the qRT-PCR experiments and preparing the manuscript. MJF and KP carried out the
statistical analysis under the supervision of NHD. AHB and KH assisted with data
analysis, experimental design, and data interpretation. KH and JRF carried out the
microarray experiments. DLB and URK supervised all aspects of the study and
preparation of the manuscript.
4.8 Acknowledgements
Supported by the National Childhood Cancer Foundation Laura and Greg Norman
Fellowship (NGG), the Children's Leukaemia and Cancer Research Foundation, Perth,
Australia and National Institutes of Health grant CA95475.
103
CHAPTER 5
THE TRITERPENOID CDDO ENHANCES DOXORUBICIN-
MEDIATED CYTOTOXICITY IN T-ALL CELLS
5.1 Abstract
Despite marked advances in survival for children with T-cell acute lymphoblastic
leukaemia (T-ALL) over the past four decades, for those patients who relapse there are
still no adequate curative therapies, and the majority of these children will succumb to
their disease. The primary reason for treatment failure is resistance to cytotoxic
chemotherapy. Novel therapies, designed to target tumour specific biologic features
are needed for treating patients with disease relapse. Using microarray gene expression
profiling, we previously demonstrated significantly different expression levels of the
anti-apoptotic gene CFLAR in specimens obtained from T-ALL patients at the time of
diagnosis that subsequently relapsed, compared to patients who remained in remission
(Gottardo et al, 2007). We hypothesised that CFLAR might be involved in treatment
failure. Consistent with our hypothesis that CFLAR expression may be up-regulated
following exposure to chemotherapy, all four paired diagnosis and relapse T-ALL
patient specimens revealed significant increases in the expression of CFLAR at relapse
(P = 0.016). We tested the novel CFLAR-inhibitor, 2-cyano-3, 12-dioxooleana-1,9
(11)-dien-28-oic acid (CDDO), in two cell lines established in our laboratory from
paediatric patients diagnosed with T-ALL. We found that CDDO displayed single
agent activity at sub-micromolar concentrations in both cell lines tested. Notably,
minimally lethal doses of CDDO resulted in significant enhancement of doxorubicin
(DOX) mediated cytotoxicity in one of the cell lines assessed. The enhanced
cytotoxicity did not appear to be related to the level of CFLAR mRNA. This study
demonstrates the potential usefulness of this novel agent in T-ALL, as an anthracycline
potentiator or anthracycline-sparing agent.
104
5.2 Introduction
The past four decades have seen markedly improved survival for children with T-cell
acute lymphoblastic leukaemia (T-ALL), which accounts for approximately 10 to 15%
of cases of ALL. Approximately 75% of patients achieve long-term disease free
survival with the use of intensive multiagent chemotherapy (Goldberg et al, 2003,
Seibel et al, 2008). However, despite these significant improvements many therapeutic
challenges remain, in particular, survival remains dismal for patients who relapse
(Goldberg et al, 2003; Einsiedel et al, 2005). Additionally, a small but significant
number of patients fail to achieve remission; this group also has a very poor prognosis
(Chessels et al, 2003). The biggest hurdle to successful treatment is resistance to
cytotoxic chemotherapy. Drug resistance may be present before the commencement of
therapy when it is termed primary or intrinsic drug resistance. Alternatively, it may be
acquired subsequent to treatment when it is termed secondary or acquired drug
resistance (reviewed in Gottesman, 2002). A major goal remains the elucidation of
molecular alterations in pathways involved in resistance to cytotoxic therapy, which
may lead to the development of novel therapies targeting specific molecules within
these deregulated pathways.
Tumour cell kill by cytotoxic chemotherapy appears to occur primarily via the
activation of apoptosis (programmed cell death) pathways (reviewed in Makin and
Hickman, 2000). Failure to induce apoptotic pathways following drug treatment seems
to be related to drug resistance (Friesen et al, 1999a; 1999b). Apoptosis occurs via two
principal mechanisms that converge on common downstream effector caspases
(caspase-3 and caspase-7). Activation of caspase-3 and caspase-7 causes cleavage of
intracellular substrates resulting in apoptotic cell death. The first apoptosis pathway is
known as the intrinsic or mitochondrial pathway, whilst the second pathway is termed
the extrinsic or death receptor mediated pathway. The intrinsic apoptosis pathway is
105
activated via DNA damage, which releases several pro-apoptotic factors, including
cytochrome c, Smac/DIABLO and apoptosis inducing factor (AIF). The release of
cytochrome c into the cytoplasm results in binding to apoptotic protease activating
factor 1 (Apaf-1), which leads to the formation of the apoptosome complex (reviewed
in Green and Reed, 1998). The caspase activator, Apaf-1, is a key element of the
apoptosome complex and binds to pro-caspase-9 resulting in the release of active
caspase-9, which in turn leads to the activation of the downstream effector caspases,
caspase-3 and caspase-7, culminating in apoptotic cell death. In contrast, the extrinsic
apoptotic pathway is initiated by ligand binding to tumour necrosis factor (TNF)-family
death receptors (so named because they posses a death-domain in their cytosolic tail
[reviewed in Locksley et al, 2001]). Examples of these receptors include TNFR1 (also
known as CD120a), Fas (also known as APO1 or CD95), and the TNF receptor
apoptosis-inducing ligand (TRAIL) death receptors DR4, DR5 and DR6. Cell surface
binding of these receptors leads to recruitment of Fas-associated death domain-
containing protein (FADD), which in turn activates mainly caspase-8, but also caspase-
10, within a complex (consisting of receptor, FADD, caspases 8 and 10) termed the
death inducing signalling complex (DISC). Once released into the cytoplasm the same
downstream effector caspases, which are involved in the intrinsic pathway, are
activated.
A unique gene expression profile according to ALL subtype, most striking between
pre-B and T-ALL, was observed in an analysis investigating the role of key apoptosis
genes in childhood ALL (Holleman et al, 2006b). Indeed, 63% (44/70) of the apoptosis
genes analysed were significantly differentially expressed between T-ALL and
pre-B ALL. In particular, many genes involved in the extrinsic pathway were
up-regulated in T-ALL, suggesting that the extrinsic pathway may play a more
significant role in T-ALL than pre-B ALL. Interestingly, Fas is predominantly
106
expressed in normal T-lymphocytes (Suda et al, 1993) and has been shown to be
involved in their apoptosis (Dhein et al, 1995). Although most cytotoxic agents are
thought to exert their cell death effects via the intrinsic pathway (reviewed in Green
and Reed, 1998; Kroemer et al, 2000; Herr et al, 2001), the extrinsic pathway also
appears to play a role in drug-induced apoptosis (Friesen et al, 1996). This led us to
speculate that the extrinsic apoptotic pathway might be deregulated in T-ALL, and
possibly play an important role in the acquisition of resistance to chemotherapy. In
keeping with this hypothesis, we previously identified CFLAR/FLIP (CASP-8 and
FADD-like apoptosis regulator), an apoptosis regulatory protein (Irmler et al, 1997)
specific for the extrinsic apoptotic pathway, as a member of a 3-gene predictor, the
expression of which, at the time of diagnosis, was able to discriminate patients with
respect to outcome (Gottardo et al, 2007). The gene encoding CFLAR/FLIP is located
on chromosome 2q33-q34 and the protein can result in blockage of the extrinsic
apoptotic pathway by binding to pro-caspase-8 and preventing cleavage to the active
caspase-8 (Irmler et al, 1997). In humans, alternative splicing results in three currently
known forms of CFLAR; long, short and Raji forms (reviewed in Yu and Shi 2008).
Both long and short isoforms function as anti-apoptotic proteins. However,
interestingly the long form has also been reported to be pro-apoptotic
(Chang et al, 2002), a function that appears to be related to the expression level of
CFLAR (reviewed in Yu and Shi 2008). Consistent with our premise that CFLAR may
play a more important role in T-ALL compared with pre-B ALL, Holleman et al
(2006b) observed significant up-regulation of CFLAR in leukaemia cells from T-ALL
patients compared with pre-B ALL patients. Furthermore, CFLAR up-regulation has
been demonstrated in the cisplatin resistant cervical cancer cell line HeLa, suggesting
the mechanism of cisplatin resistance may be the suppression of apoptosis via the
extrinsic pathway (Kamarajan et al, 2003). Additionally, in malignant mesothelioma
107
cells, increases in CFLAR protein levels have been implicated in the development of
intrinsic resistance to death receptor induced apoptosis (Rippo et al, 2004). Moreover,
TRAIL-resistant multiple myeloma cells expressing high levels of CFLAR,
demonstrated restored TRAIL sensitivity following reductions of CFLAR (Mitsiades et
al, 2002). Similarly, in B-NHL cell lines the down-regulation of CFLAR expression
level by protein synthesis inhibition, strictly correlated with restored sensitivity to
CD95-mediated apoptosis (Irisarri et al, 2000).
We hypothesised that up-regulation of CFLAR in childhood T-ALL may promote
resistance to cytotoxic agents which predominantly utilise the extrinsic apoptotic
pathway to effect cell kill. Doxorubicin (DOX), an integral drug in the treatment of
T-ALL, has been demonstrated to induce apoptosis via the extrinsic pathway
(Friesen et al, 1997). We therefore reasoned that inhibition of CFLAR would enhance
sensitivity to DOX, by “re-opening” the extrinsic apoptotic pathway. The novel agent,
2-cyano-3, 12-dioxooleana-1,9 (11)-dien-28-oic acid (CDDO), a synthetic triterpenoid
(reviewed in Liby, Yore and Sporn, 2007), has been shown to induce apoptosis in a
variety of cancer cell lines, including acute myeloid leukaemia (AML)
(Ito et al, 2000; Konopleva et al, 2002), chronic lymphocytic leukaemia (CLL)
(Pedersen et al, 2002) multiple myeloma (Chauhan et al, 2004), osteosasarcoma
(Ito et al, 2001) and breast cancer (Lapillone et al, 2003). Importantly, CDDO has been
demonstrated to inhibit CFLAR protein (Pedersen et al, 2002; Suh et al, 2003). To test
our hypothesis, we exposed two T-ALL cell lines (PER-427 and PER-604), which
express differing CFLAR mRNA levels (PER-427 expresses high levels of CFLAR,
whereas PER-604 expresses low levels) and sensitivity to DOX (Beesley et al, 2006),
to the novel agent CDDO.
108
5.3 Methods
5.3.1 Patients
Informed consent for the use of tissues for research purposes was obtained for all
individuals involved in this study according to hospital and Australian National Health
and Medical Research Council (NHMRC) guidelines. Four paired specimens were
available for study. Specimens were obtained at initial diagnosis before therapy started
and at relapse. All patients were treated on Children's Cancer Group (CCG)
risk-adjusted protocols.
5.3.2 T-ALL cell lines
We reviewed the microarray expression values of CFLAR in a panel of T-ALL cell
lines established in our laboratory from paediatric patients diagnosed with T-ALL
(n=9) and T-ALL cell lines obtained from other sources (n=6), all of which have been
previously described (Beesley et al, 2006). The T-ALL cell lines PER-427 and
PER-604 displayed the highest and lowest expression levels of CFLAR, respectively.
5.3.3 Cell culture
The cell lines PER-427 and PER-604 were cultured at 2x106/ml in RPMI-1640 medium
supplemented with 10 to 20% heat inactivated foetal calf serum,
2mM-glutamine, 10nM 2-mercaptoethanol, pyruvate and non-essential amino acids. In
addition, PER-427 cells require 300U/ml interleukin-2 for growth (Kees et al, 2003b).
Stock solutions of CDDO (0.01M) were prepared in dimethyl sulfoxide (DMSO) and
stored at –20°C. Test concentrations of CDDO and DOX (Mayne Pharma Pty Ltd, VIC
Australia) were prepared by diluting the stock solution in tissue culture medium.
109
5.3.4 Cell viability assay
The 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) assay
(Alley et al, 1988) was used for in vitro chemosensitivity testing as previously
described (Beesley et al, 2006). Briefly, 1x106 cells were seeded into each well of a
96-well plate in fresh media in the presence or absence of CDDO or DOX. Two-fold
step dilutions of each drug were performed for the following drug ranges; for DOX 1pg
ml-1 to 8µg ml-1and CDDO 11.9pM to100µM.
Culture plates were incubated for 4 days at 37°C before addition of 10µl of
filter-sterilised MTT (5mg ml-1). To dissolve formazan crystals, 100µl of acidified
isopropyl alcohol was added after 6 hours of re-incubation. Absorbance was measured
at 590nM. Each drug concentration was tested in triplicate. The drug concentration that
results in 50% cell death (IC50 value) was used as the measure of sensitivity to drug.
5.3.5 Measurement of apoptosis
We used the Annexin-V-fluorescein isothiocyanate (Annexin-V-FITC) kit
(Roche Diagnostics Corporation, Indianapolis, USA) to measure apoptotic cell death
upon exposure to drugs for 2, 4, 12, 24 and 48 hours. Labelled Annexin-V
detects phosphatidylserine (PS) exposed on the outer layer of apoptotic cells and
necrotic cells. Propidium iodide (PI) stains only the DNA of leaky necrotic cells, thus
permitting the differentiation between apoptotic and necrotic cells. The procedure was
carried out as follows: cells were stained simultaneously with FITC-conjugated
Annexin-V and PI to distinguish intact cells (Annexin-V-ve and PI-ve), early apoptotic
cells (Annexin-V+ve and PI-ve) and necrotic cells (Annexin-V+ve and PI+ve). Stained cells
were analysed by flow cytometry.
110
5.3.6 Quantitative real-time reverse transcriptase PCR
Quantitative real-time reverse transcriptase polymerase chain reaction (qRT-PCR)
(TaqMan) was employed to determine the expression level of CFLAR. The detailed
methodology has been previously described (Dallas et al, 2005). Briefly, RNA was
extracted from bone marrow specimens using the TRIZOL method (Invitrogen,
Carlsbad, CA, USA). cDNA was generated using 1µg of RNA using Omniscript
reverse transcriptase (Qiagen). CFLAR primer and probe set was obtained from
Applied Biosystems (ABI Assays on Demand, http://www.appliedbiosystems.com/).
Aliquots of total RNA extracted were used for qRT-PCR experiments according to the
manufacturer’s protocols (ABI). Serial dilutions of reference cell lines established in
our laboratory were used to generate standard curves for each target gene. The
endogenous control, ACTB, was used to normalise the expression values in each
specimen. All samples were tested in duplicate.
5.4 Results
5.4.1 CFLAR expression is higher in specimens derived from patients at relapse
compared to the paired initial diagnostic specimen
We previously showed that leukaemic cells derived from T-ALL patients who
subsequently relapsed, had significantly lower expression levels of CFLAR compared
to specimens derived from patients in continuous complete remission (CCR)
(Gottardo et al, 2007). We hypothesised that CFLAR expression may be up-regulated
following exposure to chemotherapy and thus would be expressed at higher levels at
the time of relapse. To test this we assessed the expression level of CFLAR in paired
T-ALL specimens taken at initial diagnosis and at relapse. Four paired specimens were
available for study. Consistent with our hypothesis, all four paired specimens
111
demonstrated significantly higher CFLAR expression levels at relapse compared to
initial diagnosis (P = 0.016) (Figure 5.1).
Figure 5.1 A) Level of CFLAR expression, as measured by Affymetrix microarray
gene expression analysis, at initial diagnosis (pale grey bars) and the
corresponding sample at relapse (dark grey bars). B) Box plot of the same data
(boxes indicate medians and inter-quartile range; whiskers indicate 10th and 90th
percentiles).
112
5.4.2 Increased expression of CFLAR was associated with increased DOX
resistance
To determine if there was a relationship between the level of CFLAR expression and
resistance to DOX we used the MTT assay to measure the IC50 value to DOX in high
(PER-427 cells) and low (PER-604 cells) CFLAR expressing cells. In keeping with our
hypothesis, T-ALL cells expressing high CFLAR levels (PER-427 cells) had an almost
9-fold higher (P <0.0001) IC50 value compared to the lower CFLAR expressing cell
line, PER-604 (Figure 5.2A).
5.4.3 CDDO reveals single agent cytotoxicity against T-ALL cell lines at
sub-micromolar concentrations
Next, using the 4-day MTT assay, we sought to determine whether CDDO alone affects
T-ALL cell survival. Treatment with CDDO resulted in significant cytotoxicity in both
PER-427 and PER-604 cells. The average IC50 value for CDDO for cell line PER-427
was 0.8 ± 0.055µM compared to 0.55 ± 0.102µM for PER-604 cells (P = 0.099)
(Figure 5.2B). In control cultures containing no CDDO, 100% of cells survived.
113
Figure 5.2 Cytotoxic effects of A) doxorubicin (DOX) alone and B) CDDO alone,
on T-ALL cell lines PER-427 and PER-604. A 4-day MTT assay was performed
using 1x106 cells cultured with two-fold step dilutions of DOX (dose range: 1pg ml-1 to
8µg ml-1) or CDDO (concentration range: 11.9pM to100µM). The IC50 values were
calculated and data of 3 independent experiments are presented as mean ± SEM.
5.4.4 Mechanism of cytotoxicity
To determine if CDDO cytotoxicity was due to apoptosis, PER-427 and PER-604 cells
were cultured in the presence or absence of CDDO, as a time course. Cells were treated
with two different concentrations of CDDO, which were selected for each cell line, as
determined by the MTT assays. The concentrations used were the dose corresponding
to the IC50 value (0.8µM for PER-427 and 0.55µM for PER-604) and the lowest
concentration, which resulted in maximum cell kill (3.13µM for PER-427 cells which
resulted in an average cell kill of 90% and 100µM for PER-604 cells which resulted in
an average cell kill of 80%). Cells were collected at several time points (2, 4, 12, 24
and 48 hours) and stained with Annexin-V-FITC and PI as described in the Methods
section.
At IC50 concentrations of CDDO, only minimal apoptosis was seen above baseline (no
drug) levels for both cell lines (Figure 5.3A, C). At the higher CDDO concentration, a
gradual increase in apoptosis was measured over time for PER-427 cells, with a peak of
114
51% apoptotic cells after 18 hours (Figure 5.3C). This was paralleled by an increase in
the number of necrotic cells, which was maximal (43% necrotic cells) after 48 hours
incubation (Figure 5.3D). In contrast, even using a CDDO concentration of 100µM,
only minimal apoptosis above baseline (no drug) levels, was observed in PER-604 cells
(Figure 5.3A). However, for PER-604 cells the number of necrotic cells increased
markedly after 12 hours incubation, reaching a peak of 61%, after 48 hours incubation
(Figure 5.3B).
Figure 5.3 PER-427 and PER-604 cells were cultured with CDDO at IC50
concentrations for each cell line (pale grey bars). IC50 PER-427 concentration:
0.8µM and IC50 PER-604: 0.55µM and top concentration for each cell line (dark grey
bars) see text: 3.13µM for PER-427 cells and 100µM for PER-604. A) and C) cells
were assessed for apoptosis, B) and D) for necrosis, see Methods.
115
5.4.5 Effect of CDDO on CFLAR mRNA
To assess the effect of CDDO on CFLAR mRNA levels, PER-427 and PER-604 cells
were incubated with the IC50 concentration of CDDO, individualised for each cell line
(0.8µM for PER-427 and 0.55µM for PER-604) or 0.195µM of CDDO, a sub-lethal
(for PER-427 cells) or minimally toxic concentration (for PER-604 cells), hence fore
referred to as minimal dose. CFLAR mRNA levels were measured at two time points; 2
hours (early) and 10 hours (prior to the observed apoptosis peak) after incubation with
CDDO.
After 2 hours treatment with 0.195µM CDDO, we observed an almost 15% and 20%
increase in the expression level of CFLAR in PER-427 and PER-604 cells respectively
(Figure 5.4A, C). No change in CFLAR expression from baseline was observed using
the same concentration of CDDO and 10 hours exposure for both cell lines cells
(Figure 5.4A, C). Exposure to the IC50 concentrations resulted in a similar increase in
CFLAR expression for PER-604 cells after 2 hours incubation, but no change in
expression was seen for PER-427 cells (Figure 5.4B, D). After 10 hours incubation,
CFLAR expression levels were unchanged for PER-604 cells (Figure 5.4B, D). By
contrast, for PER-427 cells, incubation for 10 hours, at the IC50 concentration of CDDO
(0.8µM) resulted in almost 20% decrease (P = 0.08) in CFLAR expression
(Figure 5.4B).
116
Figure 5.4 PER-427 cells were incubated in the presence of CDDO at A) the
minimal dose (0.195µM) and B) IC50 concentration (0.8µM). Similarly, PER-604
cells were incubated in the presence of CDDO at C) the minimal dose (0.195µM)
and D) IC50 concentration (0.55µM). CFLAR mRNA levels were measured using
qRT-PCR at 2 hours (pale grey bars) and 10 hours (dark grey bars) post incubation with
CDDO. Data is mean ± SEM for duplicate qRT-PCR measurements.
5.4.6 CDDO enhances DOX-induced cytotoxicity
We next assessed the impact on T-ALL cell survival of treatment with the minimal
dose of CDDO, in combination with DOX. Using the 4-day MTT assay, both cell lines
were treated with increasing concentrations of DOX in the absence or presence of
CDDO (0.195µM). For PER-427 cells, no significant difference was observed in the
IC50 dose for DOX in the presence or absence of CDDO (Figure 5.5A). In contrast, for
117
PER-604 cells the average DOX IC50 value decreased by almost 50% from 0.065 ±
0.005µg ml-1 to 0.031 ± 0.001µg ml-1 (P < 0.003) (Figure 5.5B).
Figure 5.5 Cytotoxic effects of DOX alone (open bars) or in combination with the
minimal dose of CDDO (0.195µM) (grey bars). A 4-day MTT assay was performed
using 1x106 cells per well. A) PER-604 cells and B) PER-427 cells were cultured in
two-fold step dilutions of DOX (dose range: 1pg ml-1 to 8µg ml-1) alone or in
combination with CDDO. The IC50 values of 3 independent experiments were
calculated and data are presented as mean ± SEM.
5.5 Discussion
Current intensive multiagent chemotherapeutic strategies fail to cure 20 to 25% of
children with T-ALL. Furthermore, for many children these intensive strategies result
in significant acute and late toxicities (Robison et al, 2003; 2005). Efforts to develop
novel treatment approaches to improve the outcome and reduce the toxicity for patients
with T-ALL are currently underway (reviewed in Pui and Jeha, 2007a). The key to
118
developing novel therapies lies in a greater understanding of T-ALL biology and
mechanisms of drug resistance.
Genomic techniques, including gene expression arrays and more recently, single-
nucleotide polymorphism (SNP) arrays have revolutionised molecular biology,
affording researchers the ability to perform high density molecular profiling of tumour
cells. Gene expression profiling has been applied to accurately identify known
prognostic subtypes of ALL (Yeoh et al, 2002) and improve prognostication
(Yeoh et al, 2002; Chiaretti et al, 2004; Gottardo et al, 2007). Additionally,
genome-wide analysis of ALL using SNP arrays has been used to uncover novel
molecular alterations leading to the disruption of key pathways involved in
leukaemogenesis (Mulligan et al, 2007).
Using microarray gene expression profiling, we previously demonstrated significantly
different expression levels of the anti-apoptotic gene CFLAR in specimens obtained at
the time of diagnosis from T-ALL patients that subsequently relapsed, compared to
patients who remained in CCR (Gottardo et al, 2007). We hypothesised that CFLAR
might be involved in treatment failure. Interestingly, CFLAR levels were found to be
lower in specimens taken from patients who subsequently relapsed. The opposite might
be expected, since CFLAR functions as an anti-apoptotic protein (Irmler et al, 1997).
However, consistent with our hypothesis that CFLAR expression may be up-regulated
following exposure to chemotherapy, all four paired diagnosis and relapse T-ALL
patient specimens examined in this study, revealed significant increases in the
expression of CFLAR at relapse. CFLAR deregulation has been implicated in the
resistance to apoptosis in numerous cancer types.
We hypothesised that CFLAR inhibition would result in “re-opening” of the extrinsic
apoptotic pathway. Our laboratory has previously revealed differential sensitivity to
119
DOX between PER-427 (high CFLAR expressing cell line) and PER-604 cells (low
CFLAR expressing cell line) (Beesley et al, 2006). In this study we confirmed the
differential sensitivity of these cell lines to DOX, PER-427 cells being almost 9-fold
more resistant. Thus, in keeping with our hypothesis, increased expression of CFLAR
was associated with increased DOX resistance. We used the triterpenoid, CDDO,
previously shown to inhibit CFLAR protein levels (Pedersen et al, 2002;
Suh et al, 2003), to test whether this agent could abrogate DOX resistance in T-ALL
cells, by restoring competence via the extrinsic apoptotic pathway. Our observation that
CDDO revealed single agent cytotoxicity to T-ALL cells, at sub-micromolar
concentrations, consistent with the findings of other groups for other leukaemic cell
lines, CLL and AML (Pedersen et al, 2002; Suh et al, 2003), prompted us to assess if
the mechanism of cytotoxicity was due to apoptosis, measured following 2, 4, 12, 24
and 48 hours exposure to drug. Incubation at IC50 concentrations of CDDO resulted in
only minimal apoptosis above background for both cell lines. Indeed, apoptosis was
observed only in PER-427 cells using a concentration 4-fold greater than the IC50
concentration of CDDO, a concentration found to result in 90% cell kill, measured
upon exposure to drugs for 4 days. In contrast, for PER-604 cells, a CDDO
concentration as high as 100µM, a concentration found to result in 80% cell kill, failed
to induce any apoptosis above background levels, but resulted in significant increases
in necrotic cell death. Our data reveals that only minimal apoptosis is induced at IC50
concentrations of CDDO in either PER-427 or PER-604 cells. However, at higher
concentrations, CDDO is capable of inducing apoptosis in certain T-ALL cells
(PER-427), but not others (PER-604). For PER-604 cells, apoptosis appears not to be
the primary mechanism of cytotoxicity, as evidenced by the marked increase in necrotic
cell death without a preceding increase in apoptotic cells. This is in contrast to Pedersen
120
et al (2002), who revealed that CDDO induced apoptosis in CLL cells by inducing the
extrinsic apoptotic pathway at micromolar concentrations.
These authors demonstrated that CDDO was effective against both sensitive and drug
resistant CLL patient specimens, by triggering the extrinsic apoptotic pathway and
demonstrated increased apoptosis in CLL cells following exposure to CDDO, which
correlated with CFLAR protein down-regulation, in a dose dependant manner
(Pedersen et al, 2002). We wondered whether CDDO might exert an inhibitory effect
on CFLAR at the mRNA level. Similar to CFLAR protein levels (Pedersen et al, 2002),
CFLAR mRNA reduction, appear to occur in a dose dependant manner. However, in
contrast to reductions in CFLAR protein levels (Pedersen et al, 2002), the degree of
CFLAR mRNA reductions induced by CDDO was only a modest 20%
(P = 0.08). Interestingly, for both T-ALL cell lines tested, CDDO appeared to induce
CFLAR mRNA expression early (2 hours of incubation with CDDO), a transient effect
that disappeared after 10 hours of incubation.
Finally, to demonstrate whether CDDO enhanced the cytotoxicity of DOX in T-ALL
cells, T-ALL cell lines expressing high and low CFLAR mRNA levels were exposed to
a combination of increasing concentrations of DOX in combination with a minimal
dose of CDDO. Cells expressing high CFLAR mRNA levels (PER-427 cells) revealed
no change in the IC50 value for DOX, demonstrating that this low dose of CDDO does
not sensitise PER-427 cells to DOX. In contrast, the DOX IC50 value was reduced
significantly by 50%, in low CFLAR expressing T-ALL cells (PER-604 cells). These
results demonstrate that in certain T-ALL cells, a minimal dose of CDDO significantly
enhances DOX-mediated cytotoxicity. In PER-604 T-ALL cells, this dose of CDDO
was not associated with a significant drop in CFLAR mRNA levels, revealing that
reduced CFLAR mRNA level is not the mechanism of action of CDDO. We speculate
that CDDO may act by inhibiting CFLAR protein, as previously shown for CLL
121
(Pedersen et al, 2002). Of note, in PER-604 cells, CDDO failed to trigger apoptosis
either at minimally lethal or at higher concentrations. Notably, Pedersen et al (2002)
demonstrated that CFLAR reductions alone were inadequate to explain the
pro-apoptotic effect of CDDO, since CFLAR knockdown using anti-sense
oligonucleotides produced less dramatic increases in apoptosis. The significance of the
modest drop in CFLAR mRNA observed using IC50 concentrations of CDDO for
PER-427 T-ALL cells remains to be determined, as we did not combine this
concentration with DOX. Perhaps enhanced DOX-mediated cytotoxicity would be
observed at higher CDDO concentrations. The observation that CDDO demonstrated
potentiation of DOX in the cell line expressing low levels of CFLAR (PER-604) was
counter-intuitive to the initial hypothesis. This may be because CDDO-mediated
cytotoxicity was via non-CFLAR related mechanisms. Future studies should include
experiments using short interfering RNA (siRNA) directed to CFLAR to investigate the
role of this gene in these cell lines.
Although anthracyclines are effective anti-leukaemic agents, their use is limited by the
potential for severe cardiac toxicity (reviewed in Kremer & Caron, 2004). Thus, CDDO
could be used to potentiate the cytotoxic effects of anthracyclines, without the need to
use potentially higher cardiotoxic doses. A potentially novel clinical application for
CDDO in the treatment of T-ALL could be to act as anthracycline-sparing agents,
minimising the dose of anthracycline required without reducing efficacy. Importantly,
in vitro studies have revealed minimal effects of CDDO on normal cells
(Chauhan et al, 2004; Ikeda et al, 2004; Kress et al, 2007). Moreover, in pre-clinical
mouse studies, no significant toxicity was observed (Lapillone et al 2003;
Konopleva et al, 2006; Kress et al, 2007). Notably, the CDDO derivative compound,
CDDO-Me, has demonstrated increased potency compared to CDDO
122
(reviewed in Liby, Yore and Sporn, 2007). The efficacy of CDDO-Me in T-ALL cells
remains to be determined.
In conclusion, our data demonstrates that the triterpenoid CDDO reveals single agent
cytotoxicity against two T-ALL cell lines at sub-micromolar concentrations. Moreover,
significant enhancement of DOX mediated cytotoxicity using only minimal doses of
CDDO can be achieved in a T-ALL cell line. The enhanced cytotoxicity did not appear
to be related to the level of CFLAR mRNA. This study demonstrates the potential
usefulness of this novel agent in T-ALL, as an anthracycline potentiator or
anthracycline-sparing agent. Additional studies using mouse models of T-ALL are
warranted to further investigate the role of this agent as a novel therapy for childhood
T-ALL.
5.7 Author contributions
NGG and URK conceived the study. The study was performed in URK’s laboratory.
NGG was responsible for carrying out all the experiments, analysing, collating, and
interpreting the data, and preparing the manuscript. JF provided the cell lines and
technical assistance. URK and DLB supervised all aspects of the study and preparation
of the manuscript.
5.8 Acknowledgements
Supported by the National Childhood Cancer Foundation Laura and Greg Norman
Fellowship (NGG), the Children’s Leukaemia and Cancer Research Foundation, Perth,
Western Australia. CDDO was kindly provided by Dr Michael Sporn of Dartmouth
College (Hanover, NH, USA).
123
CHAPTER 6
DISCUSSION
6.1 Challenges for childhood T-ALL
The treatment of childhood ALL is one of the great success stories of paediatric
oncology, transforming a universally fatal disease into one where 75 to 90% of children
are now cured (Pui & Evans, 2006a; Seibel et al, 2008). Although in the past survival
for children with T-ALL lagged behind that of children with pre-B ALL, the use of
contemporary intensified treatment strategies has significantly diminished this
difference with many groups reporting similar cure rates for both groups of patients
(Asselin et al, 2001; Schrappe et al, 2000b; Goldberg et al, 2003; Pui et al, 2004a;
Seibel et al, 2008). Intriguingly, these marked advances in survival have been achieved
through the systematic use of empirically based treatment regimens and better
understanding of established anti-leukaemic therapies rather than through increased
knowledge of the processes governing T-cell leukaemogenesis, T-lymphoblast
maintenance or the use of novel therapies. However, numerous challenges still face
physicians treating children with T-ALL. Firstly, although cure rates have significantly
improved for this group of children, there have been no additional major improvements
in outcome over the last decade, despite additional treatment intensification
(Reiter et al, 1994; Schrappe et al, 2000b; Goldberg et al, 2003; Pui et al, 2004a;
Seibel et al, 2008), suggesting that we may have reached the limit of treatment
intensification with current chemotherapeutic regimens. Secondly, effective regimens
remain elusive for treating children with relapsed T-ALL (Goldberg et al, 2003;
Einsiedel et al, 2005) or patients with resistant disease (Chessels et al, 2003). Finally,
there is a need to identify patients currently potentially overtreated and thus
unnecessarily subjected to acute and long term toxicities without benefit. A major
124
challenge therefore, is the identification of novel reliable prognostic markers, in order
to identify patients at high risk of relapse and conversely those least likely to relapse, to
guide therapy appropriately. Children predicted with a high risk of relapse would be
candidates for intensification of therapy and/or novel experimental agents. Conversely,
patients predicted to be at low risk of relapse could be offered clinical trials using
reduced intensity therapy, thereby minimising toxicity (Figure 1.2). A report from the
Berlin-Frankfurt-Münster (BFM) has provided evidence that early treatment
modification based on prednisone response results in improved patient outcome
(Schrappe et al, 2000b). To enable more precise prognostic stratification,
T-lymphoblast biology needs to be better understood. Critically, this includes the
identification of the T-ALL stem cell, the cell of origin of T-ALL, since these stem
cells may be responsible for relapse (reviewed in Reya et al, 2001;
reviewed in Warner et al, 2004).
Further treatment intensification using conventional non-specific cytotoxic therapy is
more likely to result in additional toxicity without major improvements in survival.
Intrinsic and “acquired” resistance to cytotoxic chemotherapy (discussed in chapter 5)
present the major obstacle to successful treatment. Therefore, to further improve
outcome for children with T-ALL molecular alterations in pathways involved in
resistance to cytotoxic therapy need to defined, which may permit such pathways to be
circumvented. Recent advances in molecular biology have led to a greater
understanding of the cellular changes that lead to T-cell leukaemogenesis
(reviewed in Armstrong and Look, 2005). Based on this knowledge, novel treatment
strategies will involve attempts to target molecular alterations specific to
T-lymphoblasts. Recent data suggests that the most relevant targets may actually be the
rare fraction of leukaemic stem cells (reviewed in Stubbs and Armstrong, 2007).
Current cytotoxic therapies may be effective against the bulk of T-lymphoblasts but
125
may fail to eradicate the T-ALL stem cells, ultimately resulting in relapse
(reviewed in Reya et al, 2001; and in Warner et al, 2004).
Potential hurdles to improved risk stratification are numerous. Patient outcome depends
on a complex interaction between therapy utilised, leukaemic cell biology and patient
pharmacogenetics. In addition, mechanisms leading to relapse are multifactorial and
interrelated. Although intrinsic and/or acquired drug resistance may be one of the most
important reasons for relapse, other mechanisms, including genetic polymorphisms in
chemotherapy metabolising genes have been recognised as factors affecting variability
in drug exposure and response amongst patients (Yule et al, 2001;
Masereeuw et al, 2003; Rau et al, 2006). Additionally, an individual patient’s
pharmacokinetics (Evans et al, 1998) and compliance to the treatment schedule
(Gaynon et al, 1991) also play significant roles.
The primary aim of this study was to improve on traditional prognostic markers and
develop a predictive test to be used at the time of diagnosis to determine whether an
individual patient is likely to be successfully treated on current therapy or has a high
risk of relapse. A secondary aim was to exploit the genetic differences between
relapsing and non-relapsing T-ALL patients to facilitate the development of novel
therapeutic targets.
6.2 Improving patient stratification - the genomic era
Cancer is a genetic disease. In the past, traditional molecular biology methods
permitted the study of the sequence and expression of a single gene or signalling
pathway at one time within tumour samples. Using these approaches several molecular
alterations have already been identified, some of which appear to be associated with
patient outcomes (reviewed in Armstrong and Look, 2005; reviewed in Graux et al,
126
2006). A notable recent example is the homeobox transcription factor HOX11L2, which
has been reported to be associated with poor patient outcomes (Ferrando et al, 2002;
Ballerini et al, 2002). However, another report which assessed a larger cohort of
T-ALL patients found no prognostic significance associated with HOX11L2 expression
(Cavé et al, 2004). In contrast, overexpression of the related homeobox gene HOX11 is
generally associated with a favourable prognosis in both children and adult patients
with T-ALL (Ferrando et al, 2002; 2004b; Kees et al, 2003a). The prognostic influence
of HOX11 overexpression appears to be treatment dependant. HOX11-positive T-ALL
patients treated according to Children’s Cancer Group (CCG) CCG-1901 therapy had
significantly better outcome than HOX11-negative T-ALL patients; the 5 year event
free survival (EFS) was 100% for HOX11-positive patients compared with 50% for
HOX11-negative patients (Kees et al, 2003a). On the other hand, in the same study no
difference in outcome according to HOX11 status was observed for patients treated on
the CCG-1961 protocol. These protocols employ different chemotherapeutic strategies
and the authors postulated that HOX11-positive cells may be more sensitive to certain
therapies compared with HOX11-negative cells (Kees et al, 2003a). Based on this
observation, we hypothesised that the prognostic relevance of other molecular
alterations, notably HOX11L2, may also be related to the therapy utilised. In chapter 2
the prognostic impact of HOX11L2 and HOX11 expression was assessed in a cohort of
T-ALL patients treated on CCG style therapy. Using quantitative real-time RT-PCR
(qRT-PCR), 40 samples obtained at diagnosis from children with T-ALL were
analysed. Surprisingly, and in contrast to other reports, we found that HOX11L2
overexpression was associated with a favourable outcome (Chapter 2;
Gottardo et al, 2005). Indeed, no patient whose T-lymphoblasts overexpressed
HOX11L2 suffered a relapse. Only high levels of HOX11 expression appear to confer a
prognostic advantage (Bergeron et al, 2007). Consequently, possibly due to the limited
127
number of samples expressing high levels of HOX11 (n=3) in our study, the favourable
prognosis reported to be associated with HOX11 expression was not replicated
(Chapter 2; Gottardo et al, 2005). This finding has significant clinical implications,
since such heterogeneity in expression of possible molecular prognostic markers
potentially allows T-ALL to be stratified into risk groups that predict disease behaviour
more precisely. However, before candidate molecular markers can be used to guide
treatment intensity, validation in large prospective clinical trials is required.
6.2.1 Microarray gene expression technology to identify predictive markers
The publication of the human genome sequence has afforded scientists with a view of
the location and structure of the majority of the estimated 30,000 to 40,000 human
genes and was the catalyst for the development of powerful DNA and RNA based
microarray technologies which permit the simultaneous evaluation of thousands of
genes in a tumour (reviewed in Quackenbush, 2006). DNA microarray gene expression
arrays and single nucleotide polymorphism (SNP) gene mapping arrays can measure
the expression or copy number of thousands of genes simultaneously in a tumour
sample. These approaches permit the interrogation of genes previously associated with
the biology of leukaemia, as well as uncovering promising novel underlying molecular
alterations. A myriad of tumour types have been profiled using DNA microarrays, in
both adults and children. Some of the most extensively studied to date have been the
acute leukaemias, including paediatric ALL (Ferrando et al, 2002; Ross et al, 2003;
Yeoh et al, 2002) and AML (Ross et al, 2004). These profiles have uncovered novel
molecular subtypes of ALL that provide valuable knowledge regarding the molecular
alterations that result in leukaemogenesis.
128
In order to identify a molecular signature of relapse we used DNA oligonucleotide gene
expression arrays (Affymetrix, Santa Clara, CA, USA) to characterise the gene
expression profiles of an unselected cohort of T-ALL patients treated on CCG
protocols, some of the patients later suffering a relapse. Based on these profiles we
selected a panel of predictive marker genes (n=9). The expression levels were validated
using qRT-PCR (Chapter 3; Dallas et al, 2005; Gottardo et al, 2007). Importantly, a
comprehensive analysis assessing the correlation between gene expression levels
measured by oligonucleotide microarrays and by qRT-PCR in general revealed a strong
correlation (r = 0.89) (Chapter 3; Dallas et al, 2004). Additionally, we observed a trend
towards poorer correlation for genes that exhibited fold-change differences of less than
1.5 between subsets of interest based on microarray expression scores compared to
those with fold-change differences of greater than 1.5. Our data highlight the
complementarity of oligonucleotide microarray and qRT-PCR technologies for
validation of gene expression measurements, but the poor correlation observed for 13 to
16% of genes assessed, emphasises the importance and continuing requirement for
caution in interpreting gene expression data.
Crucially, the panel of predictive marker genes that was identified by microarray
measurements was validated in a completely independent cohort of T-ALL patients,
also treated on CCG style therapy. We identified a novel set of 3 genes (CFLAR,
NOTCH2 and BTG3), termed 3-gene predictor, which distinguished patients with a
favourable outcome compared to patients with an adverse outcome (Chapter 4;
Gottardo et al, 2007). Our 3-gene predictor appears to identify a high risk group of
patients which require alternative therapeutic strategies in order to attain a cure.
Importantly, we also demonstrated that accurate outcome prediction models can be
developed using qRT-PCR expression data, a widely available technique more suitable
to analysis in the clinic. The sensitivity and specificity of the 3-gene predictor are 86%
129
and 61% respectively, whilst, the positive predictive value is 46% and the negative
predictive value is 92%. Thus, the clinical application of a stratification system such as
the one presented as part of this thesis, based on the expression level of selected genes
at diagnosis, would result in the prediction of the vast majority of patients destined to
relapse (sensitivity 86%) at the expense of the over-treatment of a significant number of
patients (specificity 61%), potentially subjecting such a group to unnecessary toxicity.
However, since relapsed T-ALL is associated with a very low chance of cure, one
could argue that this may be an acceptable compromise.
Our results also highlight that expression profiles generated using genome-wide
techniques must be interpreted cautiously, since differences among groups of interest
will frequently be observed by chance alone when such large numbers of genes are
analysed (reviewed in Simon et al, 2003). There are no statistical ways to distinguish
“real” genes from genes identified by chance alone. For example, in our analysis, four
of the seven genes, whose expression values were confirmed using qRT-PCR, were not
useful for segregating adverse and favourable outcome patients when assessed in our
independent validation cohort (Chapter 4; Gottardo et al, 2007). Thus, genes that were
highly discriminating in the training set were no longer predictive in the validation
cohort. This probably arose as a result of the small sample size of the training set, since
it is known that some of the gene signatures may be idiosyncratic to a particular
training set (Michiels et al, 2005; reviewed in Simon et al, 2003). Additionally, patients
in both the training and validation sets were treated on a variety of CCG risk-adjusted
protocols, introducing therapeutic heterogeneity as another potential source of
variation. Therefore, validation in a completely independent cohort of patients is
essential when using gene expression profiling to identify novel predictive markers.
This is especially important when the training cohort contains a small number of
samples (reviewed in Simon et al, 2003). Indeed, the proportion of misclassifications
130
has been shown to decrease as the size of the training set increased
(Michiels et al, 2005). One study assessed several non-hierarchical clustering
algorithms and found that studies comprising less than 50 samples had low
reproducibility (Garge et al, 2005).
Another criticism of using novel genetic predictive markers derived from gene
expression profiling has been a lack of validation by independent groups
(Holleman et al, 2006a). Indeed, our 3-gene predictor was not predictive in other
published data sets using in silico analysis. Since the most important prognostic
variable remains the therapy used (Nachman, 2002; Pui and Evans, 2006a), lack of
validation between independent groups may be due to the use of distinct treatment
protocols. An additional important consideration is that of therapeutic relevance when
interpreting data from retrospective studies conducted on patients who may have
received therapeutic regimens no longer in use, thus limiting the validity of results with
respect to contemporary treatment approaches. Thus, further investigation in a larger
(including at least 50 patients) and importantly uniformly treated cohort of patients is
warranted to validate our findings before clinical application of our 3-gene predictor.
Predictive gene expression signatures are now being prospectively validated in clinical
trials for patients with breast cancer (reviewed in Sotiriou and Piccart, 2007). Dutch
investigators derived a predictive signature for women with breast cancer using gene
expression profiles obtained from oligonucleotide microarrays (Agilent)
(van't Veer et al, 2002). The authors reported that outcome prediction based on a panel
of 70 genes (later named Mammoprint) more accurately predicted outcome than
stratification based on clinical variables. These investigators subsequently validated
their signature in a larger cohort of women (van de Vijver et al, 2002). Another group,
also from Holland, developed an alternative prognostic gene expression signature for
breast cancer comprised of 76 genes (Wang et al, 2005). Although these independent
131
signatures contained only a 3-gene overlap, both have subsequently been independently
validated (Buyse et al, 2006; Desmedt et al, 2007). These studies highlight the direct
clinical utility of gene expression profiles and the potential for significantly improved
patient stratification.
6.3 Towards improved outcome – identification of novel targets
Conventional chemotherapeutic agents are non-specific and cause cellular death by
damaging DNA or interfering with pathways critical for cell division. Whilst increased
understanding of tumour biology has shifted the focus onto agents that target molecular
changes necessary for leukaemogenesis and leukaemic blast cell maintenance,
amazingly, a novel cytotoxic agent which shows selective cytotoxicity for T-lineage
derived cells, including T-lymphoblasts has been developed. Nelarabine (also known as
compound 506U78, Glaxo-Wellcome), is a novel purine nucleoside, which is a soluble
pro-drug of 9-beta-D-arabinofuranosylguanine (ara-G).
Two recent phase II trials for patients with refractory or recurrent T-ALL have reported
very encouraging results using nelarabine (Berg et al, 2005; DeAngelo et al, 2007). In
the study reported by Berg et al (2005) 18 of 33 (55%) patients in first relapse had an
objective response (16 patients had a complete response (CR) and 2 had a partial
response (PR)). The objective response rate was 27% (7 CR and 1 PR) for patients in
second relapse and 33% (5 CR and 2 PR) for patients with central nervous system-
positive T-ALL or T-cell non-Hodgkin’s lymphoma (T-NHL). Whilst, only 1 out of 22
(14%) patients with extramedullary relapse had a PR.
In the adult Cancer and Leukaemia Group B, CALGB 19801 study, reported by
DeAngelo et al (2007), 39 patients with T-cell neoplasms were treated with nelarabine.
All 39 patients were evaluable for responses, which included 31% with CR and a 41%
132
overall response rate. The median overall survival (OS) was 20 weeks and the 1 year
OS 28%. Dose-limiting toxicities were central and peripheral neurotoxicity
(DeAngelo et al, 2007). Based on these results, nelarabine was granted accelerated
approval from the US Food and Drug Administration (FDA). These results are very
promising and are the subject of study by the current COG Phase III trial for T-ALL,
where nelarabine is being administered earlier in the course of disease to reduce the
incidence of relapse.
Over the past decade the remarkable progress in knowledge of the molecular biology of
tumours coupled with a need for more effective and less toxic therapies has led to the
development of a new class of anti-cancer agents. Imatinib mesylate (Gleevec,
previously called STI571, Novartis) was the first and to date the most successful of a
new class of agents targeting a specific molecular alteration, which is now termed
molecular targeted therapy. Imatinib mesylate a selective inhibitor of the constitutively
active protein tyrosine kinase coded for by BCR-ABL (Philadelphia chromosome), the
hallmark of chronic myeloid leukaemia (CML) and a subset of patients with ALL, has
significantly improved outcome for CML patients (Druker et al, 2001; 2006). The
success of imatinib mesylate in the clinic provided proof of principle that agents with
more specific activity against tumour cells than conventional cytotoxic agents could be
highly effective. This has lead to a paradigm shift in the way oncologists approach the
treatment of cancer (Druker, 2004a). Although imatinib mesylate is selective for the
BCR-ABL tyrosine kinase found in CML, it also inhibits other tyrosine kinases,
namely c-kit and platelet-derived growth factor receptors (PDGF-R). Imatinib mesylate
has been shown to be highly effective in the treatment of gastrointestinal stromal
tumors (GIFT) which express c-KIT (Blanke et al, 2008) and is also in clinical trials for
tumours which overexpress PDGF-R (reviewed in George, 2001). Notably, the finding
that the recently identified NUP214-ABL1 and EML1-ABL1 chromosomal
133
translocations appear to be sensitive to imatinib mesylate in vitro (Graux et al, 2004;
De Keersmaecker et al, 2006), provides a novel therapeutic approach for these subtypes
of T-ALL.
A myriad of small molecule inhibitors targeting genetic abnormalities and cell
signalling pathways have since been developed, many of which are currently
undergoing or have completed clinical testing. These novel agents hold great promise
to improve future therapy. However, to date only a small number of these agents have
revealed clinical utility (Druker et al, 2001; Paez et al, 2004; Lynch et al, 2004;
Vogel et al, 2002). The reasons for this may relate to the way patients are selected for
phase I and II trials. Currently, clinical trials testing the efficacy of molecularly targeted
therapies do not require that the specific molecular alteration targeted be dysregulated
in patients being treated. Therefore, such trials potentially underestimate the
effectiveness of the agent (reviewed in Druker, 2004b; reviewed in Minna et al, 2004).
The discovery that greater than 50% of T-ALL patients harbour NOTCH1 activating
mutations (Weng et al, 2004), generated great interest in using NOTCH pathway
inhibitors as novel therapies for patients with T-ALL. NOTCH1 exists in an inactive
form as a heterodimer. Binding with ligand on an adjacent cell triggers a series of
proteolytic cleavages which liberates the active NOTCH intracellular domain (ICN1),
which then translocates to the nucleus activating various target genes. The final of
these cleavages is catalysed by the gamma-secretase protein complex, which is also
responsible for the proteolysis of amyloid beta-precursor protein which causes plaque
formation in patients with Alzheimer’s disease. Gamma-secretase inhibitors were
initially developed as therapies for Alzheimer’s disease (Haass and Strooper, 1999),
however, due to their ability to also inhibit NOTCH signalling, they are appealing novel
anti-cancer agents for tumours dependent on continued NOTCH signalling. Pre-clinical
data demonstrated that T-ALL cell lines harbouring ICN1 activating mutations were
134
sensitive to gamma-secretase inhibitors (Weng et al, 2004). These inhibitors are the
first generation of NOTCH1 inhibitors to undergo clinical testing. Promising early data
from a phase I clinical trial of the gamma-secretase inhibitor, MK-0752 (Merck),
revealed activity in relapsed T-ALL (DeAngelo et al, 2006). One of four patients
harbouring a NOTCH1 activating mutation achieved a 45% reduction in a mediastinal
mass after 28 days. It is important to note that the use of agents which target critical
developmental pathways may prove to be particularly challenging in children, who are
still developing. For example, NOTCH1 plays critical roles in the development of most
organ systems including the central nervous system, heart and gastrointestinal tract
(reviewed in Artavanis-Tsakonas et al, 1999). Indeed, the dose limiting toxicity
observed using the gamma-secretase inhibitor MK-0752 was grade 3 to 4 diarrhoea
(DeAngelo et al, 2006). Other promising novel strategies for the treatment of T-ALL
include FLT3 inhibitors for the subgroup of T-ALL patients harbouring FLT3
mutations and histone deacetylase (HDAC) inhibitors for patients with TAL1
overexpression (O’Neil et al, 2004).
It is noteworthy, that a member of the 3-gene predictor identified in Chapter 4 was
another NOTCH family member, NOTCH2. Additionally, we also observed enrichment
of genes involved in the positive regulation of the NFκB pathway in patients who
subsequently relapsed (Chapter 4; Gottardo et al, 2007). The NFκB pathway has been
demonstrated to be a downstream pathway of NOTCH signalling (Vilimas et al, 2007)
and targeted disruption of this pathway with the proteosome inhibitor, bortezonib,
revealed synergism with a gamma-secretase inhibitor in T-ALL cell lines
(Vilimas et al, 2007). We hypothesised that genes associated with a relapse signature
provide promising targets for novel therapies. In chapter 5 we tested the hypothesis that
CFLAR, an inhibitor of the extrinsic apoptotic pathway and another member of the
3-gene predictor identified in Chapter 4 (Gottardo et al, 2007), may be involved in the
135
development of resistance to chemotherapy, particularly to agents which utilise the
extrinsic pathway to effect cell kill. To test our hypothesis we used a novel agent,
2-cyano-3, 12-dioxooleana-1,9 (11)-dien-28-oic acid (CDDO), previously shown to
inhibit CFLAR protein (Pedersen et al, 2002; Suh et al, 2003), in two cell lines
established in our laboratory from paediatric patients diagnosed with T-ALL. We found
that CDDO displayed single agent activity at sub-micromolar concentrations in both
cell lines tested. Importantly, minimally lethal doses of CDDO resulted in significant
enhancement of doxorubicin (DOX) mediated cytotoxicity in one of the cell lines
assessed. However, the enhanced cytotoxicity did not appear to be related to the level
of CFLAR mRNA. This study has identified a potential novel agent for the treatment of
T-ALL, which may be used as an anthracycline potentiator or anthracycline-sparing
agent.
It is critical that promising new agents be tested in pre-clinical models that best
represent the specific disease in children. To this end the National Cancer Institute
(NCI) formed the Paediatric Preclinical Testing Program (PPTP), a consortium of
investigators whose primary objective is to rapidly evaluate novel agents in pre-clinical
models (Houghton et al, 2002; 2007). The PPTP has developed an extensive panel of
paediatric tumour xenografts for the most common childhood malignancies. Data
suggests that these xenografts more accurately predict activity in paediatric clinical
trials than the NCI60 cell line panel (Johnson et al, 2001). In this way, identified novel
agents will be prioritised for clinical testing in children with relapsed or refractory
disease.
136
6.4 Future directions
Gene expression studies of leukaemic samples should become integral components of
future clinical trials. Indeed the current COG phase III trial for T-ALL is collecting
samples for analysis of gene expression profiles. Such studies should also be
accompanied by newer techniques such as SNP gene mapping arrays, which will permit
the comprehensive examination of chromosomal gains and losses. Additionally, other
novel methods, including expression analysis of microRNAs
(Cummins and Velculescu, 2006) should also be incorporated. Such studies will
provide a comprehensive molecular characterisation of T-lymphoblasts to validate the
clinical significance of known molecular abnormalities, identify molecular profiles that
will improve the accuracy of disease risk stratification and serve as targets for novel
therapies.
6.5 Summary
The findings presented as part of this thesis have revealed the value of gene expression
analysis of childhood T-ALL for identifying novel predictive markers. This study has
shown that expression profiles may provide better prognostic information than
currently available clinical variables. Additionally, genes that constitute a relapse
signature may provide rational targets for novel therapies, as demonstrated in this
study, which assessed a potential novel agent for the treatment of T-ALL.
It is anticipated that in the near future children diagnosed with T-ALL will be more
accurately stratified based on a combination of clinical variables, molecular profiles
and in vivo response to chemotherapy. Improved risk stratification will permit delivery
of individualised therapy using conventional treatment modalities in conjunction with
novel targeted therapies. Thus, patients predicted with a high risk of relapse would be
137
candidates for intensified and/or novel experimental therapies, sparing those patients
who do not require such therapy for cure. On the other hand, patients identified with a
high likelihood of cure (low risk patients) could be offered reductions in therapy, with
the benefit of reduced toxicity (Figure 1.2).
The remarkable advances made in the knowledge of tumour biology over the past
decade, which has been facilitated by genome-wide approaches, has shifted the focus
onto novel agents that target molecular changes critical for tumour proliferation or
survival. These selective agents are predicted to be less toxic to normal cells and it is
anticipated that they will be more effective than currently used non-specific
chemotherapeutic agents. The toxicity and efficacy of some of these novel agents is
currently being assessed in children with T-ALL. Ultimately, if targeted therapies prove
effective their role in combination with established anti-leukaemic agents will need to
be assessed. Importantly, many of the pathways targeted are interconnected, and
therapies targeting a single pathway or molecular lesion may not be effective. Instead,
to prevent sustained tumour proliferation, it may be necessary to utilise targeted
therapies in combination (Vilimas et al, 2007; Chan et al, 2007).
138
REFERENCES
Affymetrix technical note #1. Statistical algorithms guide.
http://www.affymetrix.com/support/technical/technotesmain.affx
Affymetrix technical note #2. Performance and validation of the GeneChip human
genome set. http://www.affymetrix.com/support/technical/technotesmain.affx
Al-Hajj M, Wicha MS, Benito-Hernandez A, Morrison SJ, Clarke MF. (2003)
Prospective identification of tumorigenic breast cancer cells. Proc Natl Acad Sci U S A.
Apr 1;100(7):3983-82003 May 27;100(11):6890.
Alley MC, Scudiero DA, Monks A, Hursey ML, Czerwinski MJ, Fine DL, Abbott BJ,
Mayo JG, Shoemaker RH, Boyd MR. (1988) Feasibility of drug screening with panels
of human tumor cell lines using a microculture tetrazolium assay.
Cancer Res. Feb 1;48(3):589-601.
Angiolillo AL, Simon C, Kodish E, Lange B, Noll RB, Ruccione K and Matloub Y.
(2004) Staged informed consent for a randomized clinical trial in childhood leukemia:
impact on the consent process. Pediatr Blood Cancer. 42: 433-437.
Aplan PD, Jones CA, Chervinsky DS, Zhao X, Ellsworth M, Wu C, McGuire EA,
Gross KW. (1997) An scl gene product lacking the transactivation domain induces
bony abnormalities and cooperates with LMO1 to generate T-cell malignancies in
transgenic mice. EMBO J. May 1;16(9):2408-19.
139
Aricò M, Valsecchi MG, Conter V, Rizzari C, Pession A, Messina C, Barisone E,
Poggi V, De Rossi G, Locatelli F, Micalizzi MC, Basso G, Masera G. (2002) Improved
outcome in high-risk childhood acute lymphoblastic leukemia defined by prednisone-
poor response treated with double Berlin-Frankfurt-Muenster protocol II.
Blood. Jul 15;100(2):420-6.
Armstrong SA, Staunton JE, Silverman LB, Pieters R, den Boer ML, Minden MD,
Sallan SE, Lander ES, Golub TR, Korsmeyer SJ. (2002) MLL translocations specify a
distinct gene expression profile that distinguishes a unique leukemia.
Nat Genet. Jan;30(1):41-7.
Armstrong SA, Look AT. (2005) Molecular genetics of acute lymphoblastic leukemia.
J Clin Oncol. Sep 10;23(26):6306-15. Review.
Artavanis-Tsakonas S, Rand MD & Lake RJ. (1999) Notch signaling: cell fate control
and signal integration in development. Science. 284, 770–776. Review.
Asnafi V, Radford-Weiss I, Dastugue N, Bayle C, Leboeuf D, Charrin C, Garand R,
Lafage-Pochitaloff M, Delabesse E, Buzyn A, Troussard X, Macintyre E. (2003)
CALM-AF10 is a common fusion transcript in T-ALL and is specific to the
TCRgammadelta lineage. Blood. Aug 1;102(3):1000-6.
Asnafi V, Beldjord K, Libura M, Villarese P, Millien C, Ballerini P, Kuhlein E, Lafage-
Pochitaloff M, Delabesse E, Bernard O, Macintyre E. (2004) Age-related phenotypic
and oncogenic differences in T-cell acute lymphoblastic leukemias may reflect thymic
atrophy. Blood. Dec 15;104(13):4173-80.
140
Asselin B, Shuster J, Amylon M, Halperin E, Hutchison R, Lipshultz S, Camitta B.
(2001) Improved Event-Free Survival (EFS) with High Dose Methotrexate (HDM) in
T-Cell Lymphoblastic Leukemia (T-ALL) and Advanced Lymphoblastic Lymphoma
(T-NHL): a Pediatric Oncology Group (POG) Study.
Proc Am Soc Clin Oncol 20: (abstr 1464).
Attarbaschi A, Mann G, Dworzak M, Wiesbauer P, Schrappe M, Gadner H. (2002)
Mediastinal mass in childhood T-cell acute lymphoblastic leukemia: significance and
therapy response. Med Pediatr Oncol. Dec;39(6):558-65.
Avramis VI. (2004) Asparaginase antibody and asparaginase activity in children with
higher-risk acute lymphoblastic leukemia. Children’s Cancer Group Study CCG-1961.
J Pediatr Hematol Oncol. 26:217-226.
Aur RJ, Simone J, Hustu HO, Walters T, Borella L, Pratt C, Pinkel D. (1971) Central
nervous system therapy and combination chemotherapy of childhood lymphocytic
leukemia. Blood. Mar;37(3):272-81.
Aur RJ, Hustu HO, Verzosa MS, Wood A, Simone JV. (1973) Comparison of two
methods of preventing central nervous system leukemia. Blood. Sep;42(3):349-57.
Bain G, Engel I, Robanus Maandag EC, te Riele HP, Voland JR, Sharp LL, Chun J,
Huey B, Pinkel D, Murre C. (1997) E2A deficiency leads to abnormalities in alphabeta
T-cell development and to rapid development of T-cell lymphomas.
Mol Cell Biol. Aug;17(8):4782-91.
141
Balis FM, Lester CM, Chrousos GP, Heideman RL, Poplack DG. (1987) Differences in
cerebrospinal fluid penetration of corticosteroids: possible relationship to the
prevention of meningeal leukemia. J Clin Oncol. Feb;5(2):202-7.
Ballerini P, Blaise A, Busson-Le Coniat M, Su XY, Zucman-Rossi J, Adam M, van den
Akker J, Perot C, Pellegrino B, Landman-Parker J, Berger R & Bernard OA. (2002)
HOX11L2 expression defines a clinical subtype of paediatric T-ALL associated with
poor prognosis. Blood. 100, 991–997.
Ballerini P, Busson M, Fasola S, van den Akker J, Lapillonne H, Romana SP, Marynen
P, Bernard OA, Landman-Parker J, Berger R. (2005) NUP214-ABL1 amplification in
t(5;14)/HOX11L2-positive ALL present with several forms and may have a prognostic
significance. Leukemia. Mar;19(3):468-70.
Barber KE, Martineau M, Harewood L, Stewart M, Cameron E, Strefford JC,
Rutherford S, Allen TD, Broadfield ZJ, Cheung KL, Harris RL, Jalali GR, Moorman
AV, Robinson HM, Harrison CJ. (2004) Amplification of the ABL gene in T-cell acute
lymphoblastic leukemia. Leukemia. Jun;18(6):1153-6.
Bash RO, Crist WM, Shuster JJ, Link MP, Amylon M, Pullen J, Carroll AJ, Buchanan
GR, Smith RG, Baer R. (1993) Clinical features and outcome of T-cell acute
lymphoblastic leukemia in childhood with respect to alterations at the TAL1 locus: a
Pediatric Oncology Group study. Blood. Apr 15;81(8):2110-7.
142
Bash RO, Hall S, Timmons CF, Crist WM, Amylon M, Smith RG, Baer R. (1995)
Does activation of the TAL1 gene occur in a majority of patients with T-cell acute
lymphoblastic leukemia? A pediatric oncology group study.
Blood. Jul 15;86(2):666-76.
Beesley, A.H., Cummings, A.J., Freitas, J.R., Hoffmann, K., Firth, M.J., Ford, J., de
Klerk, N.H. & Kees, U.R. (2005) The gene expression signature of relapse in paediatric
acute lymphoblastic leukaemia: implications for mechanisms of therapy failure.
Br J Haematol. 131, 447–456.
Beesley AH, Palmer ML, Ford J, Weller RE, Cummings AJ, Freitas JR, Firth MJ,
Perera KU, de Klerk NH, Kees UR. (2006) Authenticity and drug resistance in a panel
of acute lymphoblastic leukaemia cell lines. Br J Cancer. Dec 4;95(11):1537-44.
Begley CG, Aplan PD, Davey MP, Nakahara K, Tchorz K, Kurtzberg J, Hershfield
MS, Haynes BF, Cohen DI, Waldmann TA, Kirsch IR. (1989) Chromosomal
translocation in a human leukemic stem-cell line disrupts the T-cell antigen receptor
delta-chain diversity region and results in a previously unreported fusion transcript.
Proc Natl Acad Sci U S A. Mar;86(6):2031-5.
Bell PA, Greaves MF, Sloman JC, Thompson EN, Whittaker JA. (1983) Significance
of glucocorticoid receptors and responses in leukaemia.
J Steroid Biochem. Jul;19(1C):851-5.
143
Berg SL, Blaney SM, Devidas M, Lampkin TA, Murgo A, Bernstein M, Billett A,
Kurtzberg J, Reaman G, Gaynon P, Whitlock J, Krailo M, Harris MB; Children's
Oncology Group. (2005) Phase II study of nelarabine (compound 506U78) in children
and young adults with refractory T-cell malignancies: a report from the Children's
Oncology Group. J Clin Oncol. May 20;23(15):3376-82.
Berger R, Dastugue N, Busson M, Van Den Akker J, Pérot C, Ballerini P, Hagemeijer
A, Michaux L, Charrin C, Pages MP, Mugneret F, Andrieux J, Talmant P, Hélias C,
Mauvieux L, Lafage-Pochitaloff M, Mozziconacci MJ, Cornillet-Lefebvre P, Radford I,
Asnafi V, Bilhou-Nabera C, Nguyen Khac F, Léonard C, Speleman F, Poppe B,
Bastard C, Taviaux S, Quilichini B, Herens C, Grégoire MJ, Cavé H, Bernard OA;
Groupe Français de Cytogénétique Hématologique (GFCH). (2003) t(5;14)/HOX11L2-
positive T-cell acute lymphoblastic leukemia. A collaborative study of the Groupe
Français de Cytogénétique Hématologique (GFCH). Leukemia. Sep;17(9):1851-7.
Bergeron J, Clappier E, Cauwelier B, Dastugue N, Millien C, Delabesse E, Beldjord K,
Speleman F, Soulier J, Macintyre E, Asnafi V. (2006) HOXA cluster deregulation in
T-ALL associated with both a TCRD-HOXA and a CALM-AF10 chromosomal
translocation. Leukemia. Jun;20(6):1184-7.
Bergeron J, Clappier E, Radford I, Buzyn A, Millien C, Soler G, Ballerini P, Thomas
X, Soulier J, Dombret H, Macintyre EA, Asnafi V. (2007) Prognostic and oncogenic
relevance of TLX1/HOX11 expression level in T-ALLs. Blood. Oct 1;110(7):2324-30.
144
Bernard OA, Busson-LeConiat M, Ballerini P, Mauchauffé M, Della Valle V, Monni
R, Nguyen Khac F, Mercher T, Penard-Lacronique V, Pasturaud P, Gressin L, Heilig
R, Daniel MT, Lessard M, Berger R. (2001) A new recurrent and specific cryptic
translocation, t(5;14)(q35;q32), is associated with expression of the Hox11L2 gene in
T acute lymphoblastic leukemia. Leukemia. Oct;15(10):1495-504.
Bhojwani D, Kang H, Moskowitz NP, Min DJ, Lee H, Potter JW, Davidson G,
Willman CL, Borowitz MJ, Belitskaya-Levy I, Hunger SP, Raetz EA, Carroll WL.
(2006) Biologic pathways associated with relapse in childhood acute lymphoblastic
leukemia: a Children's Oncology Group study. Blood. Jul 15;108(2):711-7.
Blanke CD, Rankin C, Demetri GD, Ryan CW, von Mehren M, Benjamin RS,
Raymond AK, Bramwell VH, Baker LH, Maki RG, Tanaka M, Hecht JR, Heinrich
MC, Fletcher CD, Crowley JJ, Borden EC. (2008) Phase III randomized, intergroup
trial assessing imatinib mesylate at two dose levels in patients with unresectable or
metastatic gastrointestinal stromal tumors expressing the kit receptor tyrosine kinase:
S0033. J Clin Oncol. Feb 1;26(4):626-32.
Boatright KM, Deis C, Denault JB, Sutherlin DP & Salvesen GS. (2004) Activation of
caspases-8 and -10 by FLIP(L). Biochemical Journal. 382, 651–657.
Bonnet D, Dick JE. (1997) Human acute myeloid leukemia is organized as a hierarchy
that originates from a primitive hematopoietic cell. Nat Med. Jul;3(7):730-7.
Borsi JD, Moe PJ. Systemic clearance of methotrexate in the prognosis of acute
lymphoblastic leukaemia in children. (1987a) Cancer. 60:3020-3024.
145
Borsi JD, Moe PJ. (1987b) A comparative study on the pharmacokinetics of
methotrexate in a dose range of 0.5 g to 33.6 g/m2 in children with acute lymphoblastic
leukemia. Cancer. 60:5-13.
Borsi JD, Sagen E, Romslo I, Slordal L, Moe PJ. (1990) 7-Hydroxymethotrexate
concentrations in serum and cerebrospinal fluid of children with acute lymphoblastic
leukemia. Cancer Chemother Pharmacol. 27:164-167.
Bostrom BC, Sensel MR, Sather HN, Gaynon PS, La MK, Johnston K, Erdmann GR,
Gold S, Heerema NA, Hutchinson RJ, Provisor AJ, Trigg ME; Children’s Cancer
Group. (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. May 15;101(10):3809-17.
Braess J, Schneiderat P, Schoch C, Fiegl M, Lorenz I, Hiddemann W.
(2004) Functional analysis of apoptosis induction in acute myeloid leukaemia-
relevance of karyotype and clinical treatment response.
Br J Haematol. Aug; 126(3):338-47.
.
Brazma A, Hingamp P, Quackenbush J, Sherlock G, Spellman P, Stoeckert C, Aach J,
Ansorge W, Ball CA, Causton HC, Gaasterland T, Glenisson P, Holstege FC, Kim IF,
Markowitz V, Matese JC, Parkinson H, Robinson A, Sarkans U, Schulze-Kremer S,
Stewart J, Taylor R, Vilo J, Vingron M. (2001) Minimum information about a
microarray experiment (MIAME)-toward standards for microarray data.
Nat Genet. 29:365–371.
146
Breit S, Stanulla M, Flohr T, Schrappe M, Ludwig WD, Tolle G, Happich M,
Muckenthaler MU & Kulozik AE. (2006) Activating NOTCH1 mutations predict
favourable early treatment response and long-term outcome in childhood precursor
T-cell lymphoblastic leukaemia. Blood. 108, 1151–1157.
Browman GP, Goodyear MD, Levine MN, Russell R, Archibald SD, Young JE. (1990)
Modulation of the antitumor effect of methotrexate by low-dose leucovorin in
squamous cell head and neck cancer: a randomized placebo-controlled clinical trial.
J Clin Oncol. Feb;8(2):203-8. Review.
Bruce WR, Vav der Gaagh. (1963) A quantitative assay for the number of murine
lymphoma cells capable of proliferation in vitro. Nature. Jul 6;199:79-80.
Burmeister T, Gökbuget N, Reinhardt R, Rieder H, Hoelzer D, Schwartz S. (2006)
NUP214-ABL1 in adult T-ALL: the GMALL study group experience.
Blood. Nov 15;108(10):3556-9.
Buyse M, Loi S, van't Veer L, Viale G, Delorenzi M, Glas AM, d'Assignies MS, Bergh
J, Lidereau R, Ellis P, Harris A, Bogaerts J, Therasse P, Floore A, Amakrane M, Piette
F, Rutgers E, Sotiriou C, Cardoso F, Piccart MJ; TRANSBIG Consortium. (2006)
Validation and clinical utility of a 70-gene prognostic signature for women with node-
negative breast cancer. J Natl Cancer Inst. Sep 6;98(17):1183-92.
147
Cairo G, Stanulla M, Fine BM, Teuffel O, Neuhoff NV, Schrauder A, Flohr T, Schafer
BW, Bartram CR, Welte K, Schlegelberger B & Schrappe M. (2005) Distinct gene
expression profiles determine molecular treatment response in childhood acute
lymphoblastic leukaemia. Blood. 105, 821–826.
Carroll AJ, Crist WM, Link MP, Amylon MD, Pullen DJ, Ragab AH, Buchanan GR,
Wimmer RS, Vietti TJ. (2006) The t(1;14)(p34;q11) is nonrandom and restricted to
T-cell acute lymphoblastic leukemia: a Pediatric Oncology Group study.
Blood. Sep 15;76(6):1220-4.
Cauwelier B, Cavé H, Gervais C, Lessard M, Barin C, Perot C, Van den Akker J,
Mugneret F, Charrin C, Pagès MP, Grégoire MJ, Jonveaux P, Lafage-Pochitaloff M,
Mozzicconacci MJ, Terré C, Luquet I, Cornillet-Lefebvre P, Laurence B, Plessis G,
Lefebvre C, Leroux D, Antoine-Poirel H, Graux C, Mauvieux L, Heimann P, Chalas C,
Clappier E, Verhasselt B, Benoit Y, Moerloose BD, Poppe B, Van Roy N,
Keersmaecker KD, Cools J, Sigaux F, Soulier J, Hagemeijer A, Paepe AD, Dastugue
N, Berger R, Speleman F. (2007) Clinical, cytogenetic and molecular characteristics of
14 T-ALL patients carrying the TCRbeta-HOXA rearrangement: a study of the Groupe
Francophone de Cytogénétique Hématologique. Leukemia. Jan;21(1):121-8.
148
Cavé H, Suciu S, Preudhomme C, Poppe B, Robert A, Uyttebroeck A, Malet M,
Boutard, P, Benoit Y, Mauvieux L, Lutz P, Méchinaud F, Grardel N, Mazingue F,
Dupont M, Margueritte G, Pages M-P, Bertrand Y, Plouvier E, Brunie G, Bastard C,
Plantaz D, Vande Velde I, Hagemeijer A, Speleman F, Lessard M, Otten J, Vilmer E &
Dastugue N, for the EORTC-CLG. (2004) Clinical significance of HOX11L2
expression linked to t(5:14)(q35:q32), of HOX11 expression, and of SIL-TAL fusion in
childhood T-cell malignancies: results of EORTC studies 58 881 and 58 951.
Blood. 103, 442–450.
Cayuela JM, Madani A, Sanhes L, Stern MH, Sigaux F. (1996) Multiple tumor-
suppressor gene 1 inactivation is the most frequent genetic alteration in T-cell acute
lymphoblastic leukemia. Blood. Mar 15;87(6):2180-6.
Chan SM, Weng AP, Tibshirani R, Aster JC, Utz PJ. (2007) Notch signals positively
regulate activity of the mTOR pathway in T-cell acute lymphoblastic leukemia.
Blood. Jul 1;110(1):278-86.
Chang DW, Xing Z, Pan Y, Algeciras-Schimnich A, Barnhart BC, Yaish-Ohad S, Peter
ME & Yang X. (2002) c-FLIP(L) is a dual function regulator for caspase-8 activation
and CD95-mediated apoptosis. The EMBO Journal. 21, 3704–3714.
Chauhan D, Li G, Podar K, Hideshima T, Shringarpure R, Catley L, Mitsiades C,
Munshi N, Tai YT, Suh N, Gribble GW, Honda T, Schlossman R, Richardson P, Sporn
MB, Anderson KC. (2004) The bortezomib/proteasome inhibitor PS-341 and
triterpenoid CDDO-Im induce synergistic anti-multiple myeloma (MM) activity and
overcome bortezomib resistance. Blood. Apr 15;103(8):3158-66.
149
Chessells JM, Veys P, Kempski H, Henley P, Leiper A, Webb D, Hann IM. (2003)
Long-term follow-up of relapsed childhood acute lymphoblastic leukaemia.
Br J Haematol. 123:396–405.
Chiaretti S, Li X, Gentleman R, Vitale A, Vignetti M, Mandelli F, Ritz J. & Foa R.
(2004) Gene expression profile of adult T-cell acute lymphocytic leukaemia identifies
distinct subsets of patients with different response to therapy and survival.
Blood. 103, 2771–2778.
Chuaqui RF, Bonner RF, Best CJ, Gillespie JW, Flaig MJ, Hewitt SM, Phillips JL,
Krizman DB, Tangrea MA, Ahram M, Linehan WM, Knezevic V, Emmert-Buck MR.
(2002) Post-analysis follow-up and validation of microarray experiments.
Nat Genet.32:509–514.
Clappier E, Cuccuini W, Kalota A, Crinquette A, Cayuela JM, Dik WA, Langerak AW,
Montpellier B, Nadel B, Walrafen P, Delattre O, Aurias A, Leblanc T, Dombret H,
Gewirtz AM, Baruchel A, Sigaux F, Soulier J. (2007) The C-MYB locus is involved in
chromosomal translocation and genomic duplications in human T-cell acute leukemia
(T-ALL), the translocation defining a new T-ALL subtype in very young children.
Blood. Aug 15;110(4):1251-61.
Clarke M, Gaynon P, Hann I, Harrison G, Masera G, Peto R, Richards S; Childhood
ALL Collaborative Group. (2003) CNS-directed therapy for childhood acute
lymphoblastic leukemia: Childhood ALL Collaborative Group overview of 43
randomized trials. J Clin Oncol. May 1;21(9):1798-809.
150
Conter V, Aricò M, Valsecchi MG, Rizzari C, Testi AM, Messina C, Mori PG, Miniero
R, Colella R, Basso G, Rondelli R, Pession A, Masera G. (1995) Extended intrathecal
methotrexate may replace cranial irradiation for prevention of CNS relapse in children
with intermediate-risk acute lymphoblastic leukemia treated with Berlin-Frankfurt-
Münster-based intensive chemotherapy. The Associazione Italiana di Ematologia ed
Oncologia Pediatrica. J Clin Oncol. Oct;13(10):2497-502.
Conter V, Schrappe M, Aricó M, Reiter A, Rizzari C, Dördelmann M, Valsecchi MG,
Zimmermann M, Ludwig WD, Basso G, Masera G, Riehm H. (1997) Role of cranial
radiotherapy for childhood T-cell acute lymphoblastic leukemia with high WBC count
and good response to prednisone. Associazione Italiana Ematologia Oncologia
Pediatrica and the Berlin-Frankfurt-Münster groups. J Clin Oncol. Aug;15(8):2786-91.
Cummins JM, Velculescu VE. (2006) Implications of micro-RNA profiling for cancer
diagnosis. Oncogene. Oct 9;25(46):6220-7. Review.
Czechowski T, Bri RP, Stitt M, Scheible W, Udvardi MK. (2004) Real-time RT-PCR
profiling of over 1400 Arabidopsis transcription factors: unprecedented sensitivity
reveals novel root- and shoot-specific genes. Plant J.38:366–379.
Dallas PB, Gottardo NG, Firth MJ, Beesley AH, Hoffmann K, Terry PA, Freitas JR,
Boag JM, Cummings AJ & Kees UR. (2005) Gene expression levels assessed by
oligonucleotide microarray analysis and quantitative real-time RT-PCR – how well do
they correlate? BMC Genomics, 6, 59.
151
DeAngelo DJ, Stone RM, Silverman LB, Stock W, Attar EC, Fearen I, Dallob A,
Matthews C, Stone J, Freedman SJ, Aster J. (2006) Journal of Clinical Oncology, 2006
ASCO Annual Meeting Proceedings Part I. Vol 24, No. 18S (June 20 Supplement):
Abstract No 6585.
DeAngelo DJ, Yu D, Johnson JL, Coutre SE, Stone RM, Stopeck AT, Gockerman JP,
Mitchell BS, Appelbaum FR, Larson RA. (2007) Nelarabine induces complete
remissions in adults with relapsed or refractory T-lineage acute lymphoblastic leukemia
or lymphoblastic lymphoma: Cancer and Leukemia Group B study 19801.
Blood. Jun 15;109(12):5136-42.
De Keersmaecker K, Marynen P, Cools J. (2005) Genetic insights in the pathogenesis
of T-cell acute lymphoblastic leukemia. Haematologica. Aug;90(8):1116-27. Review.
De Keersmaecker K, Lahortiga I, Graux C, Marynen P, Maertens J, Cools J,
Vandenberghe P. (2006) Transition from EML1-ABL1 to NUP214-ABL1 positivity in
a patient with acute T-lymphoblastic leukemia. Leukemia. Dec;20(12):2202-4.
Desmedt C, Piette F, Loi S, Wang Y, Lallemand F, Haibe-Kains B, Viale G, Delorenzi
M, Zhang Y, d'Assignies MS, Bergh J, Lidereau R, Ellis P, Harris AL, Klijn JG,
Foekens JA, Cardoso F, Piccart MJ, Buyse M, Sotiriou C; TRANSBIG Consortium.
(2007) Strong time dependence of the 76-gene prognostic signature for node-negative
breast cancer patients in the TRANSBIG multicenter independent validation series.
Clin Cancer Res. Jun 1;13(11):3207-14.
152
Dhein J, Walczak H, Bäumler C, Debatin KM, Krammer PH. (1995) Autocrine T-cell
suicide mediated by APO-1/(Fas/CD95). Nature. Feb 2;373(6513):438-41.
Dibenedetto SP, Lo Nigro L, Mayer SP, Rovera G, Schilirò G. (1997) Detectable
molecular residual disease at the beginning of maintenance therapy indicates poor
outcome in children with T-cell acute lymphoblastic leukemia.
Blood. Aug 1;90(3):1226-32.
Dick JE. (1996) Normal and leukemic human stem cells assayed in SCID mice.
Semin Immunol. Aug;8(4):197-206. Review.
Dik WA, Brahim W, Braun C, Asnafi V, Dastugue N, Bernard OA, van Dongen JJ,
Langerak AW, Macintyre EA, Delabesse E. (2005) CALM-AF10+ T-ALL expression
profiles are characterized by overexpression of HOXA and BMI1 oncogenes.
Leukemia. Nov;19(11):1948-57.
Dowell BL, Borowitz MJ, Boyett JM, Pullen DJ, Crist WM, Quddus FF, Russell EC,
Falletta JM, Metzgar RS. (1987) Immunologic and clinicopathologic features of
common acute lymphoblastic leukemia antigen-positive childhood T-cell leukemia. A
Pediatric Oncology Group Study. Cancer. Jun 15;59(12):2020-6.
Drexler HG. (1998) Review of alterations of the cyclin-dependent kinase inhibitor
INK4 family genes p15, p16, p18 and p19 in human leukemia-lymphoma cells.
Leukemia. Jun;12(6):845-59. Review.
153
Dreyling MH, Schrader K, Fonatsch C, Schlegelberger B, Haase D, Schoch C, Ludwig
W, Löffler H, Büchner T, Wörmann B, Hiddemann W, Bohlander SK. (1998) MLL and
CALM are fused to AF10 in morphologically distinct subsets of acute leukemia with
translocation t(10;11): both rearrangements are associated with a poor prognosis.
Blood. Jun 15;91(12):4662-7.
Druker BJ, Talpaz M, Resta DJ, Peng B, Buchdunger E, Ford JM, Lydon NB,
Kantarjian H, Capdeville R, Ohno-Jones S, Sawyers CL. (2001) Efficacy and safety of
a specific inhibitor of the BCR-ABL tyrosine kinase in chronic myeloid leukemia.
N Engl J Med. Apr 5;344(14):1031-7.
Druker BJ. (2004a) Imatinib: paradigm or anomaly? Cell Cycle. Jul;3(7):833-5.
Druker BJ. (2004b) Molecularly targeted therapy: have the floodgates opened?
Oncologist. 9(4):357-60. Review.
Druker BJ, Guilhot F, O'Brien SG, Gathmann I, Kantarjian H, Gattermann N,
Deininger MW, Silver RT, Goldman JM, Stone RM, Cervantes F, Hochhaus A, Powell
BL, Gabrilove JL, Rousselot P, Reiffers J, Cornelissen JJ, Hughes T, Agis H, Fischer
T, Verhoef G, Shepherd J, Saglio G, Gratwohl A, Nielsen JL, Radich JP, Simonsson B,
Taylor K, Baccarani M, So C, Letvak L, Larson RA; IRIS Investigators. (2006)
Five-year follow-up of patients receiving imatinib for chronic myeloid leukemia.
N Engl J Med. Dec 7;355(23):2408-17.
154
Dube I D, Kamel-Reid S, Yuan C C, Lu M, Wu X, Corpus G, Raimondi S C, Crist W
M, Carroll A J, Minowanda J, Baker J B. (1991) A novel human homeobox gene lies at
the chromosome 10 breakpoint in lymphoid neoplasias with chromosomal translocation
t(10;14). Blood. 78:2996–3003.
Eden OB, Harrison G, Richards S, Lilleyman JS, Bailey CC, Chessells JM, Hann IM,
Hill FG, Gibson BE. (2000) Long-term follow-up of the United Kingdom Medical
Research Council protocols for childhood acute lymphoblastic leukaemia, 1980-1997.
Medical Research Council Childhood Leukaemia Working Party.
Leukemia. Dec;14(12):2307-20.
Einsiedel HG, von Stackelber A, Hartmann R, Fengler R, Schrappe M, Janka-Schaub
G, Mann G, Hahlen K, Gobel U, Klingebiel T, Ludwig WD & Henze G. (2005) Long-
term outcome in children with relapsed ALL by risk-stratified salvage therapy: results
of trial acute lymphoblastic leukaemia-relapse study of the Berlin-Frankfurt-Munster
Group 87. J Clin Oncol. 23, 7942–7950.
Ellisen LW, Bird J, West DC, Soreng AL, Reynolds TC, Smith SD, Sklar J. (1991)
TAN-1, the human homolog of the Drosophila notch gene, is broken by chromosomal
translocations in T lymphoblastic neoplasms. Cell. Aug 23;66(4):649-61.
Etienne W, Meyer MH, Peppers J, Meyer RA Jr. (2004) Comparison of mRNA gene
expression by RT-PCR and DNA microarray. Biotechniques.
36:618–620, 622, 624-616.
155
Evans WE, Crom WR, Stewart CF, Bowman WP, Chen CH, Abromowitch M, Simone
JV. (1984) Methotrexate systemic clearance influences probability of relapse in
children with standard-risk acute lymphocytic leukaemia. Lancet. 1:359-362.
Evans WE, Relling MV, Boyett JM, Pui CH. (1997) Does pharmacokinetic variability
influence the efficacy of high-dose methotrexate for the treatment of children with
acute lymphoblastic leukemia: what can we learn from small studies? Leuk Res.
21:435-437.
Evans WE, Relling MV, Rodman JH, Crom WR, Boyett JM, Pui CH. (1998)
Conventional compared with individualized chemotherapy for childhood acute
lymphoblastic leukemia. N Engl J Med. Feb 19;338(8):499-505.
Evans WE, Pui CH, Relling MV, III, 1999 Kluwer Academic/Plenum Publishers, New
York, 1999 pp 537- 541.
Ferrando AA, Neuberg DS, Staunton J, Loh ML, Huard C, Raimondi SC, Behm FG,
Pui CH, Downing JR, Gilliland DG, Lander ES, Golub TR & Look TL. (2002) Gene
expression signatures define novel oncogenic pathways in T cell acute lymphoblastic
leukaemia. Cancer Cell. 1, 75–87.
Ferrando AA, Armstrong SA, Neuberg DS, Sallan SE, Silverman LB, Korsmeyer SJ,
Look AT. (2003) Gene expression signatures in MLL-rearranged T-lineage and
B-precursor acute leukemias: dominance of HOX dysregulation.
Blood. Jul 1;102(1):262-8.
156
Ferrando AA, Herblot S, Palomero T, Hansen M, Hoang T, Fox EA, Look AT. (2004a)
Biallelic transcriptional activation of oncogenic transcription factors in T-cell acute
lymphoblastic leukemia. Blood. Mar 1;103(5):1909-11.
Ferrando AA, Neuberg DS, Dodge RK, Paietta E, Larson RA, Wiernik PH, Rowe JM,
Caligiuri MA, Bloomfeld CD & Look TL. (2004b) Prognostic importance of TLX1
(HOX11) oncogene expression in adults with T-cell acute lymphoblastic leukaemia.
Lancet. 363, 535–536.
Fisch P, Boehm T, Lavenir I, Larson T, Arno J, Forster A, Rabbitts TH. (1992) T-cell
acute lymphoblastic lymphoma induced in transgenic mice by the RBTN1 and RBTN2
LIM-domain genes. Oncogene. Dec;7(12):2389-97.
Friesen C, Herr I, Krammer PH, Debatin KM. (1996) Involvement of the CD95 (APO-
1/FAS) receptor/ligand system in drug-induced apoptosis in leukemia cells.
Nat Med. May;2(5):574-7.
Friesen C, Fulda S, Debatin KM. (1997) Deficient activation of the CD95 (APO-1/Fas)
system in drug-resistant cells. Leukemia. Nov;11(11):1833-41.
Friesen C, Fulda S, Debatin KM. (1999a) Cytotoxic drugs and the CD95 pathway.
Leukemia. Nov;13(11):1854-8. Review.
Friesen C, Fulda S, Debatin KM. (1999b) Induction of CD95 ligand and apoptosis by
doxorubicin is modulated by the redox state in chemosensitive- and drug-resistant
tumor cells. Cell Death Differ. May;6(5):471-80.
157
Galpin AJ, Schuetz JD, Masson E, Yanishevski Y, Synold TW, Barredo JC, Pui CH,
Relling MV, Evans WE. (1997) Differences in folylpolyglutamate synthetase and
dihydrofolate reductase expression in human B-lineage versus T-lineage leukemic
lymphoblasts: mechanisms for lineage differences in methotrexate polyglutamylation
and cytotoxicity. Mol Pharmacol. Jul;52(1):155-63.
Gardie B, Cayuela JM, Martini S, Sigaux F. (1998) Genomic alterations of the p19ARF
encoding exons in T-cell acute lymphoblastic leukemia. Blood. Feb 1;91(3):1016-20.
Garge NR, Page GP, Sprague AP, Gorman BS, Allison DB. (2005) Reproducible
clusters from microarray research: whither? BMC Bioinformatics. 2005 Jul 15;6
Gaynon PS, Bleyer WA, Steinherz PG, Finklestein JZ, Littman PS, Miller DR, Reaman
GH, Sather HN, Hammond GD. (1988a) Modified BFM therapy for children with
previously untreated acute lymphoblastic leukemia and unfavorable prognostic
features. Report of Children's Cancer Study Group Study CCG-193P.
Am J Pediatr Hematol Oncol. Spring;10(1):42-50.
Gaynon PS, Steinherz PG, Bleyer WA, Ablin AR, Albo VC, Finklestein JZ, Grossman
NJ, Littman PS, Novak LT, Pyesmany AF, et al. (1988b) Intensive therapy for children
with acute lymphoblastic leukaemia and unfavourable presenting features. Early
conclusions of study CCG-106 by the Childrens Cancer Study Group.
Lancet. Oct 22;2(8617):921-4.
158
Gaynon PS, Bleyer WA, Steinherz PG, Finklestein JZ, Littman P, Miller DR, Reaman
G, Sather H, Hammond GD. (1990) Day 7 marrow response and outcome for children
with acute lymphoblastic leukemia and unfavorable presenting features.
Med Pediatr Oncol.18(4):273-9.
Gaynon PS, Steinherz PG, Bleyer WA, Finklestein JZ, Miller DR, Reaman GH, Sather
HN, Hammond GD. (1991) Association of delivered drug dose and outcome for
children with acute lymphoblastic leukemia and unfavorable presenting features.
Med Pediatr Oncol.19(4):221-7.
Gaynon PS, Steinherz PG, Bleyer WA, Ablin AR, Albo VC, Finklestein JZ, Grossman
NJ, Novak LJ, Pyesmany AF, Reaman GH, Chappell RJ, Sather HN, Hammond GD.
(1993) Improved therapy for children with acute lymphoblastic leukemia and
unfavorable presenting features: a follow-up report of the Childrens Cancer Group
Study CCG-106. J Clin Oncol. Nov;11(11):2234-42.
Gaynon PS, Desai AA, Bostrom BC, Hutchinson RJ, Lange BJ, Nachman JB, Reaman
GH, Sather HN, Steinherz PG, Trigg ME, Tubergen DG, Uckun FM. (1997) Early
response to therapy and outcome in childhood acute lymphoblastic leukemia: a review.
Cancer. Nov 1;80(9):1717-26.
Gaynon PS, Trigg ME, Heerema NA, Sensel MG, Sather HN, Hammond GD & Bleyer
WA. (2000) Children's Cancer Group trials in childhood acute lymphoblastic
leukaemia: 1983–1995. Leukaemia. 14, 2223–2233.
159
Gaynon PS. (2005) Childhood acute lymphoblastic leukaemia and relapse.
Br J Haematol. Volume 131, Issue 5, Page 579-587.
Geley S, Hartmann BL, Hala M, Strasser-Wozak EM, Kapelari K, Kofler R. (1996)
Resistance to glucocorticoid-induced apoptosis in human T-cell acute lymphoblastic
leukemia CEM-C1 cells is due to insufficient glucocorticoid receptor expression.
Cancer Res. Nov 1;56(21):5033-8.
George D. (2001) Platelet-derived growth factor receptors: a therapeutic target in solid
tumors. Semin Oncol. Oct;28 (5 Suppl 17):27-33. Review.
Gilbertson RJ (2004) Medulloblastoma: signalling a change in treatment.
Lancet Oncol. Apr;5(4):209-18. Review.
Gilbertson RJ, Clifford SC. (2003) PDGFRB is overexpressed in metastatic
medulloblastoma. Nat Genet.35:197–198.
Gilliland DG, Griffin JD. (2002) The roles of FLT3 in hematopoiesis and leukemia.
Blood. Sep 1;100(5):1532-42. Review.
Glas AM, Floore A, Delahaye LJ, Witteveen AT, Pover RC, Bakx N, Lahti-Domenici
JS, Bruinsma TJ, Warmoes MO, Bernards R, Wessels LF, Van't Veer LJ. (2006)
Converting a breast cancer microarray signature into a high-throughput diagnostic test.
BMC Genomics. Oct 30;7:278.
160
Goldberg JM, Silverman LB, Levy DE, Dalton VK, Gelber RD, Lehmann L, Cohen
HJ, Sallan SE & Asselin BL. (2003) Childhood T-cell acute lymphoblastic leukaemia:
the Dana-Farber Cancer Institute acute lymphoblastic leukaemia consortium
experience. J Clin Oncol. 21, 3616–3622.
Goldie JH, Coldman AJ. (1979) A mathematic model for relating the drug sensitivity of
tumors to their spontaneous mutation rate. Cancer Treat Rep.
Nov-Dec;63(11-12):1727-33.
Göthert JR, Brake RL, Smeets M, Dührsen U, Begley CG, Izon DJ. (2007) NOTCH1
pathway activation is an early hallmark of SCL T leukemogenesis.
Blood. Nov 15;110(10):3753-62.
Gottardo NG, Jacoby PA, Sather HN, Reaman GH, Baker DL & Kees UR. (2005)
Significance of HOX11L2/TLX3 expression in children with T-cell acute lymphoblastic
leukaemia treated on Children's Cancer Group protocols. Leukaemia. 19, 1705–1708.
Gottesman MM. (2002) Mechanisms of cancer drug resistance.
Annu Rev Med. 53:615-27. Review.
Grabher C, von Boehmer H, Look AT. (2006) Notch 1 activation in the molecular
pathogenesis of T-cell acute lymphoblastic leukaemia. Nat Rev Cancer.
May;6(5):347-59. Review.
161
Graux C, Cools J, Melotte C, Quentmeier H, Ferrando A, Levine R, Vermeesch JR,
Stul M, Dutta B, Boeckx N, Bosly A, Heimann P, Uyttebroeck A, Mentens N, Somers
R, MacLeod RA, Drexler HG, Look AT, Gilliland DG, Michaux L, Vandenberghe P,
Wlodarska I, Marynen P, Hagemeijer A. (2004) Fusion of NUP214 to ABL1 on
amplified episomes in T-cell acute lymphoblastic leukemia.
Nat Genet. Oct;36(10):1084-9.
Graux C, Cools J, Michaux L, Vandenberghe P, Hagemeijer A. (2006) Cytogenetics
and molecular genetics of T-cell acute lymphoblastic leukemia: from thymocyte to
lymphoblast. Leukemia. Sep;20(9):1496-510. Review.
Greaves MF, Wiemels J. (2003) Origins of chromosome translocations in childhood
leukaemia. Nat Rev Cancer. Sep;3(9):639-49. Review.
Green DR, Reed JC. (1998) Mitochondria and apoptosis. Science.
Aug 28;281(5381):1309-12. Review.
Haass C, De Strooper B. (1999) The presenilins in Alzheimer's disease--proteolysis
holds the key. Science. Oct 29;286(5441):916-9. Review.
Hansen-Hagge TE, Schäfer M, Kiyoi H, Morris SW, Whitlock JA, Koch P, Bohlmann
I, Mahotka C, Bartram CR, Janssen JW. (2002) Disruption of the RanBP17/Hox11L2
region by recombination with the TCRdelta locus in acute lymphoblastic leukemias
with t(5;14)(q34;q11). Leukemia. Nov;16(11):2205-12.
162
Harrington CA, Rosenow C, Retief J. (2000) Monitoring gene expression using DNA
microarrays. Curr Opin Microbiol. Jun;3(3):285-91. Review.
Hatano M, Roberts C W M, Minden M, Crist W M, Korsmeyer S J. (1991)
Deregulation of a homeobox gene, HOX11, by the t(10;14) in T cell leukaemia.
Science. 253:79–82.
Hawley RG, Fong AZ, Reis MD, Zhang N, Lu M, Hawley TS. (1997) Transforming
function of the HOX11/TCL3 homeobox gene. Cancer Res. Jan 15;57(2):337-45.
Hayette S, Tigaud I, Maguer-Satta V, Bartholin L, Thomas X, Charrin C, Gadoux M,
Magaud JP, Rimokh R. (2002) Recurrent involvement of the MLL gene in adult
T-lineage acute lymphoblastic leukemia. Blood. Jun 15;99(12):4647-9.
Heid CA, Stevens J, Livak KJ, Williams PM. (1996) Real time quantitative PCR.
Genome Res. 6:986–994.
Heerema NA, Sather HN, Sensel MG, Kraft P, Nachman JB, Steinherz PG, Lange BJ,
Hutchinson RS, Reaman GH, Trigg ME, Arthur DC, Gaynon PS & Uckun FM. (1998)
Frequency and clinical significance of cytogenetic abnormalities in paediatric T-lineage
acute lymphoblastic leukaemia: a report from the Children's Cancer Group.
J Clin Oncol. 16, 1270–1278.
163
Heerema NA, Sather HN, Sensel MG, Lee MK, Hutchinson R, Lange BJ, Bostrom BC,
Nachman JB, Steinherz PG, Gaynon PS, Uckun FM. (2000) Clinical significance of
deletions of chromosome arm 6q in childhood acute lymphoblastic leukemia: a report
from the Children's Cancer Group. Leuk Lymphoma. Feb;36(5-6):467-78.
Heyman M, Rasool O, Borgonovo Brandter L, Liu Y, Grandér D, Söderhäll S,
Gustavsson G, Einhorn S. (1996) Prognostic importance of p15INK4B and p16INK4
gene inactivation in childhood acute lymphocytic leukemia.
J Clin Oncol. May;14(5):1512-20.
Henze G, Langermann HJ, Brämswig J, Breu H, Gadner H, Schellong G, Welte K,
Riehm H. (1981) The BFM 76/79 acute lymphoblastic leukemia therapy study
(author’s transl) Klin Padiatr. May;193(3):145-54.
Henze G, Langermann HJ, Fengler R, Brandeis M, Evers KG, Gadner H, Hinderfeld L,
Jobke A, Kornhuber B, Lampert F, Lasson U, Ludwig R, Müller-Weihrich S, Neidhardt
M, Nessler G, Niethammer D, Rister M, Ritter J, Schaaff A, Schellong G, Stollmann B,
Treuner J, Wahlen W, Weinel P, Wehinger H, Riehm H. (1982) Acute lymphoblastic
leukemia therapy study BFM 79/81 in children and adolescents: intensified reinduction
therapy for patients with different risk for relapse. Klin Padiatr.
Jul-Aug;194(4):195-203.
Herman JG, Jen J, Merlo A, Baylin SB. (1996) Hypermethylation-associated
inactivation indicates a tumor suppressor role for p15INK4B.
Cancer Res. Feb 15;56(4):722-7.
164
Herr I, Debatim KM. (2001) Cellular stress response and apoptosis in cancer therapy.
Blood. Nov 1;98(9):2603-14. Review.
Hoffmann K, Firth MJ, Freitas JR, de Klerk NH & Kees UR. (2005) Gene expression
levels in small specimens from patients detected using oligonucleotide arrays.
Molecular Biotechnology. 29, 31–38.
Hoffmann K, Firth MJ, Beesley AH, de Klerk NH & Kees UR. (2006) Translating
microarray data for diagnostic testing in childhood leukaemia. BMC Cancer. 6, 229.
Holleman A, den Boer ML, Kazemier KM, Janka-Schaub GE, Pieters R. (2003)
Resistance to different classes of drugs is associated with impaired apoptosis in
childhood acute lymphoblastic leukemia. Blood. Dec 15;102(13):4541-6.
Holleman A, Cheok MH, den Boer ML, Yang W, Veerman AJ, Kazemier KM, Pei D,
Cheng C, Pui CH, Relling MV, Janka-Schaub GE, Pieters R, Evans WE. (2004) Gene-
expression patterns in drug-resistant acute lymphoblastic leukemia cells and response
to treatment. N Engl J Med. Aug 5;351(6):533-42.
Holleman A, den Boer ML, Cheok MH, Kazemier KM, Pei D, Downing JR, Janka-
Schaub GE, Gobel U, Graubner UB, Pui CH, Evans WE, Pieters R. (2006a) Expression
of the outcome predictor in acute leukaemia 1 (OPAL1) gene is not an independent
prognostic factor in patients treated according to COALL or St Jude protocols.
Blood. 108, 1984-90.
165
Holleman A, den Boer ML, de Menezes RX, Cheok MH, Cheng C, Kazemier KM,
Janka-Schaub GE, Gobel U, Graubner UB, Evans WE & Pieters R. (2006b) The
expression of 70 apoptosis genes in relation to lineage, genetic subtype, cellular drug
resistance, and outcome in childhood acute lymphoblastic leukaemia.
Blood. 107, 769–776.
Houghton PJ, Adamson PC, Blaney S, Fine HA, Gorlick R, Haber M, Helman L,
Hirschfeld S, Hollingshead MG, Israel MA, Lock RB, Maris JM, Merlino G, Patterson
W, Reynolds CP, Shannon K, Yu A, Yu J, Smith MA. (2002) Testing of new agents in
childhood cancer preclinical models: meeting summary.
Clin Cancer Res. Dec;8(12):3646-57.
Howbrook DN, van der Valk AM, O'Shaughnessy MC, Sarker DK, Baker SC, Lloyd
AW. (2003) Developments in microarray technologies.
Drug Discov Today. 8:642–651.
Huntly BJ, Gilliland DG. (2005a) Leukaemia stem cells and the evolution of cancer-
stem-cell research. Nat Rev Cancer. Apr;5(4):311-21. Review.
Huntly BJ, Gilliland DG. (2005b) Cancer biology: summing up cancer stem cells.
Nature. Jun 30;435(7046):1169-70.
Hustu HO, Aur RJ, Verzosa MS, Simone JV, Pinkel D. (1973) Prevention of central
nervous system leukemia by irradiation. Cancer. Sep;32(3):585-97.
166
Igarashi S, Manabe A, Ohara A, Kumagai M, Saito T, Okimoto Y, Kamijo T, Isoyama
K, Kajiwara M, Sotomatsu M, Sugita K, Sugita K, Maeda M, Yabe H, Kinoshita A,
Kaneko T, Hayashi Y, Ikuta K, Hanada R, Tsuchida M. (2005) No advantage of
dexamethasone over prednisolone for the outcome of standard- and intermediate-risk
childhood acute lymphoblastic leukemia in the Tokyo Children's Cancer Study Group
L95-14 protocol.J Clin Oncol. Sep 20;23(27):6489-98.
Imamura J, Miyoshi I, Koeffler HP. (1994) p53 in hematologic malignancies.
Blood. Oct 15;84(8):2412-21. Review.
Irisarri M, Plumas J, Bonnefoix T, Jacob MC, Roucard C, Pasquier MA, Sotto JJ,
Lajmanovich A. (2000) Resistance to CD95-mediated apoptosis through constitutive
c-FLIP expression in a non-Hodgkin's lymphoma B cell line.
Leukemia. Dec;14(12):2149-58.
Irizarry RA, Hobbs B, Collin F, Beazer-Barclay YD, Antonellis KJ, Scherf U, Speed
TP. (2003a) Exploration, normalization, and summaries of high density oligonucleotide
array probe level data. Biostatistics. 4:249–264.
Irizarry RA, Bolstad BM, Collin F, Cope LM, Hobbs B, Speed TP. (2003b) Summaries
of Affymetrix GeneChip probe level data. Nucleic Acids Res. Feb 15;31(4):e15.
Irmler M, Thome M, Hahne M, Schneider P, Hofmann K, Steiner V, Bodmer JL,
Schroter M, Burns K, Mattmann C, Rimoldi D, French LE, Tschopp J. (1997)
Inhibition of death receptor signals by cellular FLIP. Nature. Jul 10;388(6638):190-5.
167
Ito Y, Pandey P, Place A, Sporn MB, Gribble GW, Honda T, Kharbanda S, Kufe D.
(2000) The novel triterpenoid 2-cyano-3,12-dioxoolean-1,9-dien-28-oic acid induces
apoptosis of human myeloid leukemia cells by a caspase-8-dependent mechanism.
Cell Growth Differ. May;11(5):261-7.
Ito Y, Pandey P, Sporn MB, Datta R, Kharbanda S, Kufe D. (2001) The novel
triterpenoid CDDO induces apoptosis and differentiation of human osteosarcoma cells
by a caspase-8 dependent mechanism. Mol Pharmacol. May;59(5):1094-9.
Jankovic M, Brouwers P, Valsecchi MG, Van Veldhuizen A, Huisman J, Kamphuis R,
Kingma A, Mor W, Van Dongen-Melman J, Ferronato L, Mancini MA, Spinetta JJ and
Masera G. (1994) Association of 1800 cGy cranial irradiation with intellectual function
in children with acute lymphoblastic leukaemia. ISPACC. International Study Group
on Psychosocial Aspects of Childhood Cancer. Lancet. Jul 23;344(8917):224-7.
Johnson JI, Decker S, Zaharevitz D, Rubinstein LV, Venditti JM, Schepartz S,
Kalyandrug S, Christian M, Arbuck S, Hollingshead M, Sausville EA. (2001)
Relationships between drug activity in NCI preclinical in vitro and in vivo models and
early clinical trials. Br J Cancer. May 18;84(10):1424-31.
Jordan B. (2002) Historical background and anticipated developments.
Ann N Y Acad Sci. Dec; 975:24–32. Review.
168
Kager L, Cheok M, Yang W, Zaza G, Cheng Q, Panetta JC, Pui CH, Downing JR,
Relling MV, Evans WE. (2005) Folate pathway gene expression differs in subtypes of
acute lymphoblastic leukemia and influences methotrexate pharmacodynamics.
J Clin Invest. Jan;115(1):110-7
Kamarajan P, Sun NK, Chao CC. (2003) Up-regulation of FLIP in cisplatin-selected
HeLa cells causes cross-resistance to CD95/Fas death signalling.
Biochem J. Nov 15;376 (Pt 1):253-60.
Kaspers GJ, Veerman AJ, Popp-Snijders C, Lomecky M, Van Zantwijk CH, Swinkels
LM, Van Wering ER, Pieters R. (1996) Comparison of the antileukemic activity in
vitro of dexamethasone and prednisolone in childhood acute lymphoblastic leukemia.
Med Pediatr Oncol. Aug;27(2):114-21.
Kaspers GJ, Wijnands JJ, Hartmann R, Huismans L, Loonen AH, Stackelberg A,
Henze G, Pieters R, Hählen K, Van Wering ER, Veerman AJ. (2005)
Immunophenotypic cell lineage and in vitro cellular drug resistance in childhood
relapsed acute lymphoblastic leukaemia. Eur J Cancer. Jun;41(9):1300-3.
Kato GJ, Quddus FF, Shuster JJ, Boyett J, Pullen JD, Borowitz MJ, Whitehead VM,
Crist WM, Leventhal BG. (1993) High glucocorticoid receptor content of leukemic
blasts is a favourable prognostic factor in childhood acute lymphoblastic leukemia.
Blood. Oct 15;82(8):2304-9.
169
Kawamura-Tsuzuku J, Suzuki T, Yoshida Y & Yamamoto T. (2004) Nuclear
localization of Tob is important for regulation of its antiproliferative activity.
Oncogene. 23, 6630–6638.
Kees UR, Burton PR, Lü C, Baker DL. (1997) Homozygous deletion of the p16/MTS1
gene in pediatric acute lymphoblastic leukemia is associated with unfavorable clinical
outcome. Blood. Jun 1;89(11):4161-6.
Kees UR, Carter TL, Watt PM, Kumar R, Baker DL, Reaman GH, Sather HN, Burton
PR. (2001) p16INK4A gene deletion in pediatric acute lymphoblastic leukemia. Blood.
97:4003–4004.
Kees UR, Heerema NA, Kumar R, Watt PM, Baker DL, La MK, Uckun FM & Sather
HN. (2003a) Expression of HOX11 in childhood T-lineage acute lymphoblastic
leukaemia can occur in the absence of cytogenetic aberration at 10q24: a study from the
Children's Cancer Group (CCG). Leukaemia. 17, 887–893.
Kees UR, Ford J, Watson M, Murch A, Ringner M, Walker RL, Meltzer P (2003b)
Gene expression profiles in a panel of childhood leukemia cell lines mirror critical
features of the disease. Mol Cancer Ther 2: 671–677.
Kees UR. (2004) Gene expression signatures in lymphoid tumours.
Immunol Cell Biol. 82:154–160.
Kelly PN, Dakic A, Adams JM, Nutt SL, Strasser A. (2007) Tumor growth need not be
driven by rare cancer stem cells. Science. Jul 20;317(5836):337.
170
Kennedy MA, Gonzalez-Sarmiento R, Kees UR, Lampert F, Dear N, Boehm T,
Rabbitts TH. (1991) HOX11, a homeobox-containing T-cell oncogene on human
chromosome 10q24. Proc Natl Acad Sci U S A. Oct 15;88(20):8900-4.
Klein D. (2002) Quantification using real-time PCR technology: applications and
limitations. Trends Mol Med. 8:257–260.
Kmita M, Duboule D. (2003) Organizing axes in time and space; 25 years of colinear
tinkering. Science. Jul 18;301(5631):331-3. Review.
Konopleva M, Tsao T, Ruvolo P, Stiouf I, Estrov Z, Leysath CE, Zhao S, Harris D,
Chang S, Jackson CE, Munsell M, Suh N, Gribble G, Honda T, May WS, Sporn MB,
Andreeff M. (2002) Novel triterpenoid CDDO-Me is a potent inducer of apoptosis and
differentiation in acute myelogenous leukemia. Blood. Jan 1;99(1):326-35.
Konopleva M, Zhang W, Shi YX, McQueen T, Tsao T, Abdelrahim M, Munsell MF,
Johansen M, Yu D, Madden T, Safe SH, Hung MC, Andreeff M. (2006) Synthetic
triterpenoid 2-cyano-3,12-dioxooleana-1,9-dien-28-oic acid induces growth arrest in
HER2-overexpressing breast cancer cells. Mol Cancer Ther. Feb;5(2):317-28.
Kordes U, Krappmann D, Heissmeyer V, Ludwig WD & Scheidereit C. (2000)
Transcription factor NF-kappaB is constitutively activated in acute lymphoblastic
leukaemia cells. Leukaemia. 3, 399–402.
Kremer LC, Caron HN. (2004) Anthracycline cardiotoxicity in children.
N Engl J Med. Jul 8;351(2):120-1.
171
Kress CL, Konopleva M, Martinez-Garcia V, Krajewska M, Lefebvre S, Hyer ML,
McQueen T, Andreeff M, Reed JC, Zapata JM. (2007) Triterpenoids display single
agent anti-tumor activity in a transgenic mouse model of chronic lymphocytic leukemia
and small B cell lymphoma. PLoS ONE. Jun 27;2(6):e559.
Kroemer G, Reed JC. (2000) Mitochondrial control of cell death.
Nat Med. May;6(5):513-9. Review.
Lahortiga I, De Keersmaecker K, Van Vlierberghe P, Graux C, Cauwelier B, Lambert
F, Mentens N, Beverloo HB, Pieters R, Speleman F, Odero MD, Bauters M, Froyen G,
Marynen P, Vandenberghe P, Wlodarska I, Meijerink JP, Cools J. (2007) Duplication
of the MYB oncogene in T cell acute lymphoblastic leukemia.
Nat Genet. May;39(5):593-5.
Lapillonne H, Konopleva M, Tsao T, Gold D, McQueen T, Sutherland RL, Madden T,
Andreeff M. (2003) Activation of peroxisome proliferator-activated receptor gamma by
a novel synthetic triterpenoid 2-cyano-3,12-dioxooleana-1,9-dien-28-oic acid induces
growth arrest and apoptosis in breast cancer cells. Cancer Res. Sep 15;63(18):5926-39.
Larson RC, Fisch P, Larson TA, Lavenir I, Langford T, King G, Rabbitts TH. (1994) T
cell tumours of disparate phenotype in mice transgenic for Rbtn-2. Oncogene.
Dec;9(12):3675-81.
172
Larson RC, Lavenir I, Larson TA, Baer R, Warren AJ, Wadman I, Nottage K, Rabbitts
TH. (1996) Protein dimerization between Lmo2 (Rbtn2) and Tal1 alters thymocyte
development and potentiates T cell tumorigenesis in transgenic mice.
EMBO J. Mar 1;15(5):1021-7.
Laughton SJ, Ashton LJ, Kwan E, Norris MD, Haber M, Marshall GM. (2005) Early
responses to chemotherapy of normal and malignant hematologic cells are prognostic in
children with acute lymphoblastic leukemia. J Clin Oncol. Apr1;23(10):2264-71.
le Viseur C, Hotfilder M, Bomken S, Wilson K, Röttgers S, Schrauder A, Rosemann A,
Irving J, Stam RW, Shultz LD, Harbott J, Jürgens H, Schrappe M, Pieters R, Vormoor
J. In childhood acute lymphoblastic leukemia, blasts at different stages of
immunophenotypic maturation have stem cell properties. Cancer Cell.
2008 Jul 8;14(1):47-58
Liby KT, Yore MM, Sporn MB. (2007) Triterpenoids and rexinoids as multifunctional
agents for the prevention and treatment of cancer. Nat Rev Cancer.
May;7(5):357-69. Review.
Lichty BD, Ackland-Snow J, Noble L, Kamel-Reid S, Dubé ID. (1995) Dysregulation
of HOX11 by chromosome translocations in T-cell acute lymphoblastic leukemia: a
paradigm for homeobox gene involvement in human cancer.
Leuk Lymphoma. Jan;16(3-4):209-15. Review.
173
Liu G, Loraine AE, Shigeta R, Cline M, Cheng J, Valmeekam V, Sun S, Kulp D, Siani-
Rose MA. (2003) NetAffx: Affymetrix probesets and annotations.
Nucl Acids Res. 31:82–86.
Livak KJ, Flood SJ, Marmaro J, Giusti W, Deetz K. (1995) Oligonucleotides with
fluorescent dyes at opposite ends provide a quenched probe system useful for detecting
PCR product and nucleic acid hybridization. PCR Methods Appl. 4:357–362.
Locksley RM, Killeen N, Lenardo MJ. (2001) The TNF and TNF receptor
superfamilies: integrating mammalian biology. Cell. Feb 23;104(4):487-501. Review.
Lu M, Gong Z, Shen W, Ho A D. (1991) The tcl-3 proto-oncogene altered by
chromosomal translocation in T-cell leukaemia codes for a homeobox protein.
EMBO J. 10:2905–2910.
Ludwig WD, Thiel E, Bartram CR, Kranz BR, Raghavachar A, Löffler H, Ganser A,
Büchner T, Hiddemann W, Heil G, et al. (1990) Clinical importance of T-ALL
subclassification according to thymic or prethymic maturation stage.
Haematol Blood Transfus. 33:419-27.
Lugthart S, Cheok MH, den Boer ML, Yang W, Holleman A, Cheng C, Pui CH,
Relling, MV, Janka-Schaub GE, Pieters R & Evans WE. (2005) Identification of genes
associated with chemotherapy crossresistance and treatment response in childhood
acute lymphoblastic leukaemia. Cancer Cell. 7, 375–386.
174
Lynch TJ, Bell DW, Sordella R, Gurubhagavatula S, Okimoto RA, Brannigan BW,
Harris PL, Haserlat SM, Supko JG, Haluska FG, Louis DN, Christiani DC, Settleman J,
Haber DA. (2004) Activating mutations in the epidermal growth factor receptor
underlying responsiveness of non-small-cell lung cancer to gefitinib.
N Engl J Med. May 20;350(21):2129-39.
McGuire EA, Hockett RD, Pollock KM, Bartholdi MF, O'Brien SJ, Korsmeyer SJ.
(1989) The t(11;14)(p15;q11) in a T-cell acute lymphoblastic leukemia cell line
activates multiple transcripts, including Ttg-1, a gene encoding a potential zinc finger
protein. Mol Cell Biol. May;9(5):2124-32.
McGuire EA, Rintoul CE, Sclar GM, Korsmeyer SJ. (1992) Thymic overexpression of
Ttg-1 in transgenic mice results in T-cell acute lymphoblastic leukemia/lymphoma.
Mol Cell Biol. Sep;12(9):4186-96.
Maloney KW, Shuster JJ, Murphy S, Pullen J, Camitta BA. (2000) Long-term results of
treatment studies for childhood acute lymphoblastic leukemia: Pediatric Oncology
Group studies from 1986-1994. Leukemia. Dec;14(12):2276-85.
Makin G, Hickman JA. (2000) Apoptosis and cancer chemotherapy.
Cell Tissue Res. Jul;301(1):143-52. Review.
Marshall GM, Haber M, Kwan E, Zhu L, Ferrara D, Xue C, Brisco MJ, Sykes PJ,
Morley A, Webster B, Dalla Pozza L, Waters K, Norris MD. (2003) Importance of
minimal residual disease testing during the second year of therapy for children with
acute lymphoblastic leukemia. J Clin Oncol. Feb 15;21(4):704-9.
175
Masereeuw R, Notenboom S, Smeets PH, Wouterse AC, Russel FG. (2003) Impaired
renal secretion of substrates for the multidrug resistance protein 2 in mutant transport-
deficient (TR-) rats. J Am Soc Nephrol. 14:2741-2749.
Masson E, Relling MV, Synold TW, Liu Q, Schuetz JD, Sandlund JT, Pui CH, Evans
WE. (1996) Accumulation of methotrexate polyglutamates in lymphoblasts is a
determinant of antileukemic effects in vivo. A rationale for high-dose methotrexate.
J Clin Invest. Jan 1;97(1):73-80.
Matherly LH, Taub JW, Wong SC, Simpson PM, Ekizian R, Buck S, Williamson M,
Amylon M, Pullen J, Camitta B, Ravindranath Y. (1997) Increased frequency of
expression of elevated dihydrofolate reductase in T-cell versus B-precursor acute
lymphoblastic leukemia in children. Blood. Jul 15;90(2):578-89.
Matsuda S, Rouault J, Magaud J & Berthet C. (2001) In search of a function for the
TIS21/PC3/BTG1/TOB family. FEBS Letters. 497, 67–72 (Review).
Mattano LA Jr, Sather HN, Trigg ME, Nachman JB. (2000) Osteonecrosis as a
complication of treating acute lymphoblastic leukemia in children: a report from the
Children’s Cancer Group. J Clin Oncol. Sep 15;18(18):3262-72.
Mattano LA, Sather HN, La MK, Nachman JB & Seibel NL. (2003) Modified
dexamethasone (DXM) reduces the incidence of treatment-related osteonecrosis (ON)
in children and adolescents with higher risk acute lymphoblastic leukemia (HR ALL): a
report of CCG-1961. Blood, 102, 221a.
176
Mellentin JD, Smith SD, Cleary ML. (1989) lyl-1, a novel gene altered by
chromosomal translocation in T cell leukemia, codes for a protein with a helix-loop-
helix DNA binding motif. Cell. Jul 14;58(1):77-83.
Merlo A, Herman JG, Mao L, Lee DJ, Gabrielson E, Burger PC, Baylin SB, Sidransky
D. (1995) 5' CpG island methylation is associated with transcriptional silencing of the
tumour suppressor p16/CDKN2/MTS1 in human cancers. Nat Med. Jul;1(7):686-92.
Merup M, Moreno TC, Heyman M, Rönnberg K, Grandér D, Detlofsson R, Rasool O,
Liu Y, Söderhäll S, Juliusson G, Gahrton G, Einhorn S. (1998) 6q deletions in acute
lymphoblastic leukemia and non-Hodgkin's lymphomas. Blood. May 1;91(9):3397-400.
Michiels S, Koscielny S, Hill C. (2005) Prediction of cancer outcome with microarrays:
a multiple random validation strategy. Lancet. Feb 5-11;365(9458):488-92.
Miller LD, Long PM, Wong L, Mukherjee S, McShane LM, Liu ET. (2002) Optimal
gene expression analysis by microarrays. Cancer Cell. Nov;2(5):353-61. Review.
Minna JD, Gazdar AF, Sprang SR, Herz J. (2004) Cancer. A bull's eye for targeted lung
cancer therapy. Science. Jun 4;304(5676):1458-61.
Mitchell CD, Richards SM, Kinsey SE, Lilleyman J, Vora A, Eden TO; Medical
Research Council Childhood Leukaemia Working Party. (2005) Benefit of
dexamethasone compared with prednisolone for childhood acute lymphoblastic
leukaemia: results of the UK Medical Research Council ALL97 randomized trial.
Br J Haematol. Jun;129(6):734-45.
177
Mitsiades N, Mitsiades CS, Poulaki V, Anderson KC, Treon SP. (2002) Intracellular
regulation of tumor necrosis factor-related apoptosis-inducing ligand-induced apoptosis
in human multiple myeloma cells. Blood. Mar 15;99(6):2162-71.
Moghrabi A, Levy DE, Asselin B, Barr R, Clavell L, Hurwitz C, Samson Y, Schorin
M, Dalton VK, Lipshultz SE, Neuberg DS, Gelber RD, Cohen HJ, Sallan SE,
Silverman LB. (2007) Results of the Dana-Farber Cancer Institute ALL Consortium
Protocol 95-01. Results of the Dana-Farber Cancer Institute ALL Consortium Protocol
95-01 for children with acute lymphoblastic leukemia. Blood. Feb 1;109(3):896-904.
Mocellin S, Rossi CR, Pilati P, Nitti D, Marincola FM. (2003) Quantitative real-time
PCR: a powerful ally in cancer research. Trends Mol Med. 9:189–195.
Mook S, Van't Veer LJ, Rutgers EJ, Piccart-Gebhart MJ, Cardoso F. (2007)
Individualization of therapy using Mammaprint: from development to the MINDACT
Trial. Cancer Genomics Proteomics. May-Jun;4(3):147-55. Review.
Moreau Y, Aerts S, De Moor B, De Strooper B, Dabrowski M. (2003) Comparison and
meta-analysis of microarray data: from the bench to the computer desk. Trends Genet.
19:570–577.
Mosquera-Caro M, Helman P, Veroff R, Shuster J, Martin S, Davidson G, Potter J,
Harvey R, Hromas R, Andries E, Atlas S, Wilson C, Ar K, Xu Y, Chen IM, Carroll A,
Camitta B & Willman C. (2003) Identification, validation and cloning of a novel gene
(OPAL1) and associated genes highly predictive of outcome in paediatric acute
lymphoblastic leukaemia using gene expression profiling [abstract]. Blood. 102, 4a.
178
Mulhern RK, Fairclough D, Ochs J. (1991) A prospective comparison of
neuropsychologic performance of children surviving leukemia who received 18-Gy,
24-Gy, or no cranial irradiation. J Clin Oncol. Aug;9(8):1348-56.
Mullighan CG, Flotho C, Downing JR. (2005) Genomic assessment of pediatric acute
leukemia. Cancer J. Jul-Aug;11(4):268-82. Review.
Mullighan CG, Goorha S, Radtke I, Miller CB, Coustan-Smith E, Dalton JD, Girtman
K, Mathew S, Ma J, Pounds SB, Su X, Pui CH, Relling MV, Evans WE, Shurtleff SA,
Downing JR. (2007) Genome-wide analysis of genetic alterations in acute
lymphoblastic leukaemia. Nature. Apr 12;446(7137):758-64.
Mutch DM, Berger A, Mansourian R, Rytz A, Roberts MA. (2002) The limit fold
change model: a practical approach for selecting differentially expressed genes from
microarray data. BMC Bioinformatics. 3:17.
Nachman J, Sather HN, Gaynon PS, Lukens JN, Wolff L, Trigg ME. (1997)
Augmented Berlin-Frankfurt-Münster therapy abrogates the adverse prognostic
significance of slow early response to induction chemotherapy for children and
adolescents with acute lymphoblastic leukaemia and favourable presenting features: a
report from the Children’s Cancer Group. J Clin Oncol. Jun;15(6):2222-30.
Nachman JB, Sather HN, Sensel MG, Trigg ME, Cherlow JM, Lukens JN, Wolff L,
Uckun FM & Gaynon PS. (1998) Augmented post-induction therapy for children with
high-risk acute lymphoblastic leukaemia and a slow response to initial therapy.
N Engl J Med. 338, 1663–1671.
179
Nachman JB. (2002) Treatment: the most important prognostic variable. J Pediatr
Hematol Oncol. Dec; 24(9):704-5.
Nachman J. (2005) Clinical characteristics, biologic features and outcome for young
adult patients with acute lymphoblastic leukaemia. Br J Haematol.
Jul;130(2):166-73. Review.
Nathan PC, Maze R, Spiegler B, Greenberg ML, Weitzman S, Hitzler JK. (2004) CNS-
directed therapy in young children with T-lineage acute lymphoblastic leukemia: High-
dose methotrexate versus cranial irradiation. Pediatr Blood Cancer. Jan;42(1):24-9.
Neale GA, Menarguez J, Kitchingman GR, Fitzgerald TJ, Koehler M, Mirro J Jr,
Goorha RM. (1991) Detection of minimal residual disease in T-cell acute
lymphoblastic leukemia using polymerase chain reaction predicts impending relapse.
Blood. Aug 1;78(3):739-47.
Neale GA, Rehg JE, Goorha RM. (1995) Ectopic expression of rhombotin-2 causes
selective expansion of CD4-CD8- lymphocytes in the thymus and T-cell tumors in
transgenic mice. Blood. Oct 15;86(8):3060-71.
Niehues T, Kapaun P, Harms DO, Burdach S, Kramm C, Körholz D, Janka-Schaub G,
Göbel U. (1999) A classification based on T cell selection-related phenotypes identifies
a subgroup of childhood T-ALL with favourable outcome in the COALL studies.
Leukemia. Apr;13(4):614-7.
180
Olivier M, Eeles R, Hollstein M, Khan MA, Harris CC, Hainaut P. (2002) The IARC
TP53 database: new online mutation analysis and recommendations to users. Hum
Mutat. Jun;19(6):607-14.
O'Neil J, Billa M, Oikemus S, Kelliher M. (2001) The DNA binding activity of TAL-1
is not required to induce leukemia/lymphoma in mice. Oncogene. Jun 28;20(29):3897-
905.
O'Neil J, Shank J, Cusson N, Murre C, Kelliher M. (2004) TAL1/SCL induces
leukemia by inhibiting the transcriptional activity of E47/HEB.
Cancer Cell. Jun;5(6):587-96.
O'Neil J, Grim J, Strack P, Rao S, Tibbitts D, Winter C, Hardwick J, Welcker M,
Meijerink JP, Pieters R, Draetta G, Sears R, Clurman BE, Look AT. (2007) FBW7
mutations in leukemic cells mediate NOTCH pathway activation and resistance to
gamma-secretase inhibitors. J Exp Med. Aug 6;204(8):1813-24.
O'Neil J, Look AT. (2007) Mechanisms of transcription factor deregulation in
lymphoid cell transformation. Oncogene. Oct 15;26(47):6838-49. Review.
Packer RJ, Meadows AT, Rorke LB, Goldwein JL, D'Angio G. (1987) Long-term
sequelae of cancer treatment on the central nervous system in childhood.
Med Pediatr Oncol. 15(5):241-53. Review.
181
Paez JG, Jänne PA, Lee JC, Tracy S, Greulich H, Gabriel S, Herman P, Kaye FJ,
Lindeman N, Boggon TJ, Naoki K, Sasaki H, Fujii Y, Eck MJ, Sellers WR, Johnson
BE, Meyerson M. (2004) EGFR mutations in lung cancer: correlation with clinical
response to gefitinib therapy. Science. Jun 4;304(5676):1497-500.
Paietta E, Ferrando AA, Neuberg D, Bennett JM, Racevskis J, Lazarus H, Dewald G,
Rowe JM, Wiernik PH, Tallman MS, Look AT. (2004) Activating FLT3 mutations in
CD117/KIT(+) T-cell acute lymphoblastic leukemias. Blood. Jul 15;104(2):558-60.
Palomero T, Lim WK, Odom DT, Sulis ML, Real PJ, Margolin A, Barnes KC, O'Neil
J, Neuberg D, Weng AP, Aster JC, Sigaux F, Soulier J, Look AT, Young RA, Califano
A, Ferrando AA. (2006) NOTCH1 directly regulates c-MYC and activates a feed-
forward-loop transcriptional network promoting leukemic cell growth.
Proc Natl Acad Sci U S A. Nov 28;103(48):18261-6.
Palomero T, Sulis ML, Cortina M, Real PJ, Barnes K, Ciofani M, Caparros E, Buteau
J, Brown K, Perkins SL, Bhagat G, Agarwal AM, Basso G, Castillo M, Nagase S,
Cordon-Cardo C, Parsons R, Zúñiga-Pflücker JC, Dominguez M, Ferrando AA. (2007)
Mutational loss of PTEN induces resistance to NOTCH1 inhibition in T-cell leukemia.
Nat Med. Oct;13(10):1203-10.
Pardal R, Clarke MF, Morrison SJ. (2003) Applying the principles of stem-cell biology
to cancer. Nat Rev Cancer. Dec;3(12):895-902. Review.
182
Parkin D, Whelan S, Ferlay J, Teppo L, Thomas D, eds. (2003) Cancer incidence in
five continents, vol. VIII (IARC Scientific Publication No. 155). Lyon, France:
International Agency for Research on Cancer.
Parr C, Watkins G & Jiang WG. (2004) The possible correlation of Notch-1 and Notch-
2 with clinical outcome and tumour clinicopathological parameters in human breast
cancer. International Journal of Molecular Medicine. 5, 779–786.
Pear WS, Aster JC, Scott ML, Hasserjian RP, Soffer B, Sklar J, Baltimore D. (1996)
Exclusive development of T cell neoplasms in mice transplanted with BM expressing
activated Notch alleles. J Exp Med. May 1; 183(5):2283-91.
Pear WS, Aster JC. (2004) T cell acute lymphoblastic leukemia/lymphoma: a human
cancer commonly associated with aberrant NOTCH1 signaling. Curr Opin Hematol.
Nov;11(6):426-33. Review.
Pedersen IM, Kitada S, Schimmer A, Kim Y, Zapata JM, Charboneau L, Rassenti L,
Andreeff M, Bennett F, Sporn MB, Liotta LD, Kipps TJ, Reed JC. (2002) The
triterpenoid CDDO induces apoptosis in refractory CLL B cells.
Blood. Oct 15;100(8):2965-72.
Petricoin EF 3rd, Hackett JL, Lesko LJ, Puri RK, Gutman SI, Chumakov K, Woodcock
J, Feigal DW Jr, Zoon KC, Sistare FD. (2002) Medical applications of microarray
technologies: a regulatory science perspective. Nat Genet. 32:474–479.
183
Pieters R, Kaspers GJ, van Wering ER, Huismans DR, Loonen AH, Hählen K,
Veerman AJ. (1993) Cellular drug resistance profiles that might explain the prognostic
value of immunophenotype and age in childhood acute lymphoblastic leukemia.
Leukemia. Mar;7(3):392-7.
Pieters R, den Boer ML, Durian M, Janka G, Schmiegelow K, Kaspers GJ, van Wering
ER, Veerman AJ. (1998) Relation between age, immunophenotype and in vitro drug
resistance in 395 children with acute lymphoblastic leukemia--implications for
treatment of infants. Leukemia. Sep;12(9):1344-8.
Pui CH, Behm FG, Singh B, Schell MJ, Williams DL, Rivera GK, Kalwinsky DK,
Sandlund JT, Crist WM, Raimondi SC. (1990) Heterogeneity of presenting features and
their relation to treatment outcome in 120 children with T-cell acute lymphoblastic
leukemia. Blood. Jan 1;75(1):174-9.
Pui CH, Sallan S, Relling MV, Masera G, Evans WE. (2001) International Childhood
Acute Lymphoblastic Leukemia Workshop: Sausalito, CA, 30 November-1 December
2000. Leukemia. May;15(5):707-15.
Pui CH, Cheng C, Leung W, Rai SN, Rivera GK, Sandlund JT, Ribeiro RC, Relling
MV, Kun LE, Evans WE, Hudson MM. (2003) Extended follow-up of long-term
survivors of childhood acute lymphoblastic leukemia. N Engl J Med.
Aug 14;349(7):640-9.
184
Pui CH, Sandlund JT, Pei D, Campana D, Rivera GK, Ribeiro RC, Rubnitz JE,
Razzouk BI, Howard SC, Hudson MM, Cheng C, Kun LE, Raimondi SC, Behm FG,
Downing JR, Relling MV, Evans WE; Total Therapy Study XIIIB at St Jude Children’s
Research Hospital. (2004a) Improved outcome for children with acute lymphoblastic
leukemia: results of Total Therapy Study XIIIB at St Jude Children’s Research
Hospital. Blood. Nov 1;104(9):2690-6.
Pui CH, Relling MV, Downing JR. (2004b) Acute lymphoblastic leukemia.
N Engl J Med. Apr 8;350(15):1535-48. Review.
Pui CH. (2005) Quest for effective agents to combat T-cell acute lymphoblastic
leukemia. Eur J Cancer. Jun;41(9):1243-5.
Pui CH, & Evans WE. (2006a) Treatment of acute lymphoblastic leukaemia.
N Engl J Med. 354, 166–178. Review.
Pui CH, Relling MV, Evans WE. (2006b) Is mega dose of methotrexate beneficial to
patients with acute lymphoblastic leukemia? Leuk Lymphoma. 47:2431-2432.
Pui CH. (2006c) Central nervous system disease in acute lymphoblastic leukemia:
prophylaxis and treatment. Hematology Am Soc Hematol Educ Program. 142–146.
Pui CH, Jeha S. (2007a) New therapeutic strategies for the treatment of acute
lymphoblastic leukaemia. Nat Rev Drug Discov. Feb;6(2):149-65. Review.
185
Pui CH, Campana D. (2007b) Age-related differences in leukemia biology and
prognosis: the paradigm of MLL-AF4-positive acute lymphoblastic leukemia.
Leukemia. Apr;21(4):593-4.
Pullen J, Boyett J, Shuster J, Crist W, Land V, Frankel L, Iyer R, Backstrom L, van Eys
J, Harris M, Ravindranath Y, Sullivan M. (1993) Extended triple intrathecal
chemotherapy trial for prevention of CNS relapse in good-risk and poor-risk patients
with B-progenitor acute lymphoblastic leukemia: a Pediatric Oncology Group study.
J Clin Oncol. May;11(5):839-49.
Pullen J, Shuster JJ, Link M, Borowitz M, Amylon M, Carroll AJ, Land V, Look AT,
McIntyre B & Camitta B. (1999) Significance of commonly used prognostic factors
differs for children with T cell acute lymphocytic leukaemia (ALL), as compared to
those with B-precursor ALL. A Paediatric Oncology Group (POG) study. Leukaemia.
11, 1696–1707.
Quackenbush J. (2006) Microarray analysis and tumor classification.
N Engl J Med. 354:2463-2472.
Quelle DE, Zindy F, Ashmun RA, Sherr CJ. (1995) Alternative reading frames of the
INK4a tumor suppressor gene encode two unrelated proteins capable of inducing cell
cycle arrest. Cell. Dec 15;83(6):993-1000.
Rabbitts TH. (1991) Translocations, master genes, and differences between the origins
of acute and chronic leukemias. Cell. Nov 15;67(4):641-4. Review.
186
Rabbitts TH. (1993) Ciba Medal Lecture. Chromosomal abnormalities in cancer cells:
identification of molecules important for tumour development.
Biochem Soc Trans. Nov;21(4):809-15. Review.
Rabbitts TH. (1998) LMO T-cell translocation oncogenes typify genes activated by
chromosomal translocations that alter transcription and developmental processes.
Genes Dev. Sep 1;12(17):2651-7. Review.
Rau T, Erney B, Gores R, Eschenhagen T, Beck J, Langer T. (2006) High-dose
methotrexate in pediatric acute lymphoblastic leukemia: impact of ABCC2
polymorphisms on plasma concentrations. Clin Pharmacol Ther. 80:468-476.
Reiter A, Schrappe M, Ludwig WD, Tiemann M, Parwaresch R, Zimmermann M,
Schirg E, Henze G, Schellong G, Gadner H, Riehm H. (2000) Intensive ALL-type
therapy without local radiotherapy provides a 90% event-free survival for children with
T-cell lymphoblastic lymphoma: a BFM group report. Blood. Jan 15;95(2):416-21.
Ramakers-van Woerden NL, Pieters R, Slater RM, Loonen AH, Beverloo HB, van
Drunen E, Heyman M, Moreno TC, Rots MG, van Wering ER, Kamps WA, Janka-
Schaub GE, Veerman AJ. (2001) In vitro drug resistance and prognostic impact of
p16INK4A/P15INK4B deletions in childhood T-cell acute lymphoblastic leukaemia.
Br J Haematol. Mar;112(3):680-90.
Reya T, Morrison SJ, Clarke MF, Weissman IL. (2001) Stem cells, cancer, and cancer
stem cells. Nature. Nov 1;414(6859):105-11. Review.
187
Rippo MR, Moretti S, Vescovi S, Tomasetti M, Orecchia S, Amici G, Catalano A,
Procopio A. (2004) FLIP overexpression inhibits death receptor-induced apoptosis in
malignant mesothelial cells. Oncogene. Oct 14;23(47):7753-60.
Rivera GK, Pinkel D, Simone JV, Hancock ML, Crist WM. (1993) Treatment of acute
lymphoblastic leukemia. 30 years' experience at St. Jude Children's Research Hospital.
N Engl J Med. Oct 28;329(18):1289-95.
Robb L, Lyons I, Li R, Hartley L, Köntgen F, Harvey RP, Metcalf D, Begley CG.
(1995) Absence of yolk sac hematopoiesis from mice with a targeted disruption of the
scl gene. Proc Natl Acad Sci U S A. Jul 18;92(15):7075-9.
Robison LL, Bhatia S. (2003) Late-effects among survivors of leukaemia and
lymphoma during childhood and adolescence. Br J Haematol. Aug;122(3):345-59.
Review.
Robison LL, Green DM, Hudson M, Meadows AT, Mertens AC, Packer RJ, Sklar CA,
Strong LC, Yasui Y, Zeltzer LK. (2005) Long-term outcomes of adult survivors of
childhood cancer. Cancer. Dec 1;104(11 Suppl):2557-64.
Rockett JC, Hellmann GM. (2004) Confirming microarray data – is it really necessary?
Genomics. 83:541–549.
188
Ross ME, Zhou X, Song G, Shurtleff SA, Girtman K, Williams WK, Liu HC, Mahfouz
R, Raimondi SC, Lenny N, Patel A, Downing JR. (2003) Classification of pediatric
acute lymphoblastic leukemia by gene expression profiling.
Blood. Oct 15;102(8):2951-9.
Ross ME, Mahfouz R, Onciu M, Liu HC, Zhou X, Song G, Shurtleff SA, Pounds S,
Cheng C, Ma J, Ribeiro RC, Rubnitz JE, Girtman K, Williams WK, Raimondi SC,
Liang DC, Shih LY, Pui CH, Downing JR. (2004) Gene expression profiling of
pediatric acute myelogenous leukemia. Blood. Dec 1;104(12):3679-87.
Royer-Pokora B, Loos U, Ludwig WD. (1991) TTG-2, a new gene encoding a cysteine-
rich protein with the LIM motif, is overexpressed in acute T-cell leukaemia with the
t(11;14)(p13;q11). Oncogene. Oct;6(10):1887-93.
Roussel MF. (1999) The INK4 family of cell cycle inhibitors in cancer.
Oncogene. Sep 20;18(38):5311-7. Review.
Rubnitz JE, Behm FG, Pui CH, Evans WE, Relling MV, Raimondi SC, Harrison PL,
Sandlund JT, Ribeiro RC, Grosveld G, Downing JR. (1997) Genetic studies of
childhood acute lymphoblastic leukemia with emphasis on p16, MLL, and ETV6 gene
abnormalities: results of St Jude Total Therapy Study XII.
Leukemia. Aug;11(8):1201-6.
189
Rubnitz JE, Camitta BM, Mahmoud H, Raimondi SC, Carroll AJ, Borowitz MJ,
Shuster JJ, Link MP, Pullen DJ, Downing JR, Behm FG, Pui CH. (1999) Childhood
acute lymphoblastic leukemia with the MLL-ENL fusion and t(11;19)(q23;p13.3)
translocation. J Clin Oncol. Jan;17(1):191-6.
Russo G, Zegar C, Giordano A. (2003) Advantages and limitations of microarray
technology in human cancer. Oncogene. 22:6497–6507.
Salvati PD, Ranford PR, Ford J, Kees UR. (1995) HOX11 expression in pediatric acute
lymphoblastic leukemia is associated with T-cell phenotype.
Oncogene. Oct 5;11(7):1333-8.
Salvati PD, Watt PM, Thomas WR, Kees UR. (1999) Molecular characterization of a
complex chromosomal translocation breakpoint t(10;14) including the HOX11
oncogene locus. Leukemia. Jun;13(6):975-9.
Schneider NR Carroll AJ, Shuster JJ, Pullen J, Link MP, Borowitz MJ, Camitta BM,
Katz JA and Amylon MD. (2000) New recurring cytogenetic abnormalities and
associations of blast cell karyotypes with prognosis in childhood T-cell acute
lymphoblastic leukemia: a Pediatric Oncology Group report of 343 cases.
Blood. 96:2543-2549.
Schrappe M, Reiter A, Zimmermann M, Harbott J, Ludwig WD, Henze G, Gadner H,
Odenwald E, Riehm H. (2000a) Long-term results of four consecutive trials in
childhood ALL performed by the ALL-BFM study group from 1981 to 1995.
Berlin-Frankfurt-Münster. Leukemia. Dec;14(12):2205-22.
190
Schrappe M, Reiter A, Ludwig WD, Harbott J, Zimmermann M, Hiddemann W,
Niemeyer C, Henze G, Feldges A, Zintl F, Kornhuber B, Ritter J, Welte K, Gadner H,
Riehm H. (2000b) 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. Jun 1;95(11):3310-22.
Schultz KR, Pullen DJ, Sather HN, Shuster JJ, Devidas M, Borowitz MJ, Carroll AJ,
Heerema NA, Rubnitz JE, Loh ML, Raetz EA, Winick NJ, Hunger SP, Carroll WL,
Gaynon PS, Camitta BM. (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. Feb 1;109(3):926-35.
Seibel NL, Asselin BL, Nachman JB, Steinherz P, Camitta B, Devidas M & Sather HN.
(2004) Treatment of high risk T-cell acute lymphoblastic leukemia (T-ALL):
comparison of recent experience of the Children's Cancer Group (CCG) and Pediatric
Oncology Group (POG) [abstract]. Blood. 104, 681.
Seibel NL, Steinherz PG, Sather HN, Nachman JB, Delaat C, Ettinger LJ, Freyer DR,
Mattano LA Jr, Hastings CA, Rubin CM, Bertolone K, Franklin JL, Heerema NA,
Mitchell TL, Pyesmany AF, La MK, Edens C, Gaynon PS. (2008) Early post-induction
intensification therapy improves survival for children and adolescents with high-risk
acute lymphoblastic leukemia: a report from the Children's Oncology Group.
Blood. Mar 1;111(5):2548-55.
191
Seidel H, Nygaard R, Moe PJ, Jacobsen G, Lindqvist B, Slordal L. (1997) On the
prognostic value of systemic methotrexate clearance in childhood acute lymphocytic
leukemia. Leuk Res. 21:429-434.
Sharma VM, Calvo JA, Draheim KM, Cunningham LA, Hermance N, Beverly L,
Krishnamoorthy V, Bhasin M, Capobianco AJ, Kelliher MA. (2006) Notch1
contributes to mouse T-cell leukemia by directly inducing the expression of c-myc.
Mol Cell Biol. Nov;26(21):8022-31.
Sharpless NE, DePinho RA. (1999) The INK4A/ARF locus and its two gene products.
Curr Opin Genet Dev. Feb;9(1):22-30. Review.
Shirasawa S, Arata A, Onimaru H, Roth KA, Brown GA, Horning S, Arata S, Okumura
K, Sasazuki T, Korsmeyer SJ. (2000) Rnx deficiency results in congenital central
hypoventilation. Nat Genet. Mar;24(3):287-90.
Shivdasani RA, Mayer EL, Orkin SH. (1995) Absence of blood formation in mice
lacking the T-cell leukaemia oncoprotein tal-1/SCL. Nature. Feb 2;373(6513):432-4.
Sherlock, G. (2005) Of fish and chips. Nature Methods. 2, 329–330.
Shipp MA, Ross KN, Tamayo P, Weng AP, Kutok JL, Aguiar RC, Gaasenbeek M,
Angelo M, Reich M, Pinkus GS, Ray TS, Koval MA, Last KW, Norton A, Lister TA,
Mesirov J, Neuberg DS, Lander ES, Aster JC, Golub TR. (2002) Diffuse large B-cell
lymphoma outcome prediction by gene-expression profiling and supervised machine
learning. Nat Med. 2002 Jan;8(1):68-74.
192
Shuster JJ, Falletta JM, Pullen DJ, Crist WM, Humphrey GB, Dowell BL, Wharam
MD, Borowitz M. (1990) Prognostic factors in childhood T-cell acute lymphoblastic
leukemia: a Pediatric Oncology Group study. Blood. Jan 1;75(1):166-73.
Silverman LB, Declerck L, Gelber RD, Dalton VK, Asselin BL, Barr RD, Clavell LA,
Hurwitz CA, Moghrabi A, Samson Y, Schorin MA, Lipton JM, Cohen HJ, Sallan SE.
(2000) Results of Dana-Farber Cancer Institute Consortium protocols for children with
newly diagnosed acute lymphoblastic leukemia (1981-1995).
Leukemia. Dec;14(12):2247-56.
Simon R, Radmacher MD, Dobbin K, McShane LM (2003) Pitfalls in the use of DNA
microarray data for diagnostic and prognostic classification. J Natl Cancer Inst.
Jan 1;95(1):14-8. Review.
Simone JV. (2003) Childhood leukemia--successes and challenges for survivors.
N Engl J Med. Aug 14;349(7):627-8.
Sinclair PB, Sorour A, Martineau M, Harrison CJ, Mitchell WA, O'Neill E, Foroni L.
(2004) A fluorescence in situ hybridization map of 6q deletions in acute lymphocytic
leukemia: identification and analysis of a candidate tumor suppressor gene.
Cancer Res. Jun 15;64(12):4089-98.
Singh SK, Hawkins C, Clarke ID, Squire JA, Bayani J, Hide T, Henkelman RM,
Cusimano MD, Dirks PB. (2004) Identification of human brain tumour initiating cells.
Nature. 2004 Nov 18;432(7015):396-401.
193
Sjölund J, Manetopoulos C, Stockhausen MT, Axelson H. (2005) The Notch pathway
in cancer: differentiation gone awry. Eur J Cancer. Nov;41(17):2620-9. Review.
Soulier J, Clappier E, Cayuela JM, Regnault A, García-Peydró M, Dombret H,
Baruchel A, Toribio ML, Sigaux F. (2005) HOXA genes are included in genetic and
biologic networks defining human acute T-cell leukemia (T-ALL).
Blood. Jul 1;106(1):274-86.
Speleman F, Cauwelier B, Dastugue N, Cools J, Verhasselt B, Poppe B, Van Roy N,
Vandesompele J, Graux C, Uyttebroeck A, Boogaerts M, De Moerloose B, Benoit Y,
Selleslag D, Billiet J, Robert A, Huguet F, Vandenberghe P, De Paepe A, Marynen P,
Hagemeijer A. (2005) A new recurrent inversion, inv(7)(p15q34), leads to
transcriptional activation of HOXA10 and HOXA11 in a subset of T-cell acute
lymphoblastic leukemias. Leukemia. Mar;19(3):358-66.
Steinherz PG, Gaynon P, Miller DR, Reaman G, Bleyer A, Finklestein J, Evans RG,
Meyers P, Steinherz LJ, Sather H, Hammond D. (1986) Improved disease-free survival
of children with acute lymphoblastic leukaemia at high risk for early relapse with the
New York regimen - a new intensive therapy protocol: a report from the Childrens
Cancer Study Group. J Clin Oncol. 4:744-752.
Steinherz PG, Gaynon PS, Breneman JC, Cherlow JM, Grossman NJ, Kersey JH,
Johnstone HS, Sather HN, Trigg ME, Uckun FM and Bleyer WA. (1998) Treatment of
patients with acute lymphoblastic leukaemia with bulky extramedullary disease and
T-cell phenotype and other poor prognostic features. Cancer. 3: 600-612.
194
Stubbs MC, Armstrong SA. (2007) Therapeutic implications of leukemia stem cell
development. Clin Cancer Res. Jun 15;13(12):3439-42. Review.
Su XY, Busson M, Della Valle V, Ballerini P, Dastugue N, Talmant P, Ferrando AA,
Baudry-Bluteau D, Romana S, Berger R, Bernard OA. (2004) Various types of
rearrangements target TLX3 locus in T-cell acute lymphoblastic leukemia.
Genes Chromosomes Cancer. Nov;41(3):243-9.
Suda T, Takahashi T, Golstein P, Nagata S. (1993) Molecular cloning and expression
of the Fas ligand, a novel member of the tumor necrosis factor family.
Cell. Dec 17;75(6):1169-78.
Suh WS, Kim YS, Schimmer AD, Kitada S, Minden M, Andreeff M, Suh N, Sporn M,
Reed JC. (2003) Synthetic triterpenoids activate a pathway for apoptosis in AML cells
involving downregulation of FLIP and sensitization to TRAIL.
Leukemia. Nov;17(11):2122-9.
Takeuchi S, Bartram CR, Seriu T, Miller CW, Tobler A, Janssen JW, Reiter A, Ludwig
WD, Zimmermann M, Schwaller J, Lee E, Miyoshi I, Koeffler HP. (1995) Analysis of
a family of cyclin-dependent kinase inhibitors: p15/MTS2/INK4B, p16/MTS1/INK4A,
and p18 genes in acute lymphoblastic leukemia of childhood.
Blood. Jul 15;86(2):755-60.
195
Takeuchi S, Koike M, Seriu T, Bartram CR, Schrappe M, Reiter A, Park S, Taub HE,
Kubonishi I, Miyoshi I, Koeffler HP. (1998) Frequent loss of heterozygosity on the
long arm of chromosome 6: identification of two distinct regions of deletion in
childhood acute lymphoblastic leukemia. Cancer Res. Jun 15;58(12):2618-23.
Thompson BJ, Buonamici S, Sulis ML, Palomero T, Vilimas T, Basso G, Ferrando A,
Aifantis I. (2007). The SCFFBW7 ubiquitin ligase complex as a tumor suppressor in
T cell leukemia. J Exp Med. Aug 6;204(8):1825-35.
Tsuchida M, Ikuta K, Hanada R, Saito T, Isoyama K, Sugita K, Toyoda Y, Manabe A,
Koike K, Kinoshita A, Maeda M, Ishimoto K, Sato T, Okimoto Y, Kaneko T, Kajiwara
M, Sotomatsu M, Hayashi Y, Yabe H, Hosoya R, Hoshi Y, Ohira M, Bessho F,
Tsunematsu Y, Tsukimoto I, Nakazawa S. (2000) Long-term follow-up of childhood
acute lymphoblastic leukemia in Tokyo Children's Cancer Study Group 1981-1995.
Leukemia. Dec;14(12):2295-306.
Uckun FM, Reaman G, Steinherez PG, Arthur DC, Sather H, Trigg M, Tubergen D,
Gaynon P. (1996a) Improved clinical outcome for children with T-lineage acute
lymphoblastic leukemia after contemporary chemotherapy: a Children's Cancer Group
Study. Leuk Lymphoma. Dec;24(1-2):57-70.
Uckun FM, Steinherz PG, Sather H, Trigg M, Arthur D, Tubergen D, Gaynon P,
Reaman G. (1996b) CD2 antigen expression on leukemic cells as a predictor of event-
free survival after chemotherapy for T-lineage acute lymphoblastic leukemia: a
Children's Cancer Group study. Blood. Dec 1;88(11):4288-95.
196
Uckun FM, Gaynon PS, Sensel MG, Nachman J, Trigg ME, Steinherz PG, Hutchinson
R, Bostrom BC, Sather HN, Reaman GH. (1997) Clinical features and treatment
outcome of childhood T-lineage acute lymphoblastic leukemia according to the
apparent maturational stage of T-lineage leukemic blasts: a Children's Cancer Group
study. J Clin Oncol. Jun;15(6):2214-21.
Uckun FM, Sensel MG, Sun L, Steinherz PG, Trigg ME, Heerema NA, Sather HN,
Reaman GH, Gaynon PS. (1998) Biology and treatment of childhood T-lineage acute
lymphoblastic leukemia. Blood. Feb 1; 91(3):735-46. Review.
Valente G, Manfroi F, Peracchio C, Nicotra G, Castino R, Nicosia G, Kerim S &
Isidoro C. (2006) cFLIP expression correlates with tumour progression and patient
outcome in non-Hodgkin lymphomas of low grade of malignancy.
Br J Haematol. 132, 560–570.
Valnet-Rabier MB, Challier B, Thiebault S, Angonin R, Margueritte G, Mougin C,
Kantelip B, Deconinck E, Cahn JY & Fest T. (2005) c-Flip protein expression in
Burkitt's lymphomas is associated with a poor clinical outcome.
Br J Haematol. 128, 767–773.
van Dongen JJ, Seriu T, Panzer-Grümayer ER, Biondi A, Pongers-Willemse MJ, Corral
L, Stolz F, Schrappe M, Masera G, Kamps WA, Gadner H, van Wering ER, Ludwig
WD, Basso G, de Bruijn MA, Cazzaniga G, Hettinger K, van der Does-van den Berg A,
Hop WC, Riehm H, Bartram CR. (1998) Prognostic value of minimal residual disease
in acute lymphoblastic leukaemia in childhood. Lancet. Nov 28;352(9142):1731-8.
197
van Grotel M, Meijerink JP, Beverloo HB, Langerak AW, Buys-Gladdines JG,
Schneider P, Poulsen TS, den Boer ML, Horstmann M, Kamps WA, Veerman AJ, van
Wering ER, van Noesel MM, Pieters R. (2006) The outcome of molecular-cytogenetic
subgroups in pediatric T-cell acute lymphoblastic leukemia: a retrospective study of
patients treated according to DCOG or COALL protocols.
Haematologica. Sep;91(9):1212-21.
van Grotel M, Meijerink JP, van Wering ER, Langerak AW, Beverloo HB, Buijs-
Gladdines JG, Burger NB, Passier M, van Lieshout EM, Kamps WA, Veerman AJ, van
Noesel MM, Pieters R. (2008) Prognostic significance of molecular-cytogenetic
abnormalities in pediatric T-ALL is not explained by immunophenotypic differences.
Leukemia. Jan;22(1):124-31.
van 't Veer LJ, Dai H, van de Vijver MJ, He YD, Hart AA, Mao M, Peterse HL, van
der Kooy K, Marton MJ, Witteveen AT, Schreiber GJ, Kerkhoven RM, Roberts C,
Linsley PS, Bernards R, Friend SH. (2002) Gene expression profiling predicts clinical
outcome of breast cancer. Nature. Jan 31;415(6871):530-6.
Van Vlierberghe P, Meijerink JP, Stam RW, van der Smissen W, van Wering ER,
Beverloo HB, Pieters R. (2005) Activating FLT3 mutations in CD4+/CD8- pediatric
T-cell acute lymphoblastic leukemias. Blood. Dec 15;106(13):4414-5.
Vilimas T, Mascarenhas J, Palomero T, Mandal M, Buonamici S, Meng F, Thompson
B, Spaulding C, Macaroun S, Alegre ML, Kee BL, Ferrando A, Miele L, Aifantis I.
(2007) Targeting the NF-kappaB signaling pathway in Notch1-induced T-cell
leukemia. Nat Med. Jan;13(1):70-7.
198
Vilmer E, Suciu S, Ferster A, Bertrand Y, Cave H, Thyss A, Benoit Y, Dastugue N,
Fournier M, Souillet G, Manel AM, Robert A, Nelken B, Millot F, Lutz P, Rialland X,
Mechinaud F, Boutard P, Behar C, Chantraine JM, Plouvier E, Laureys G, Brock P,
Uyttebroeck A, Margueritte G, Plantaz D, Norton L, Francotte N, Gyselinck J,
Waterkeyn C, Solbu G, Philippe N, Otten J. (2000) Long-term results of three
randomized trials (58831, 58832, 58881) in childhood acute lymphoblastic leukemia: a
CLCG-EORTC report. Children leukaemia Cooperative Group.
Leukemia. Dec; 14(12):2257-66.
Vogel CL, Cobleigh MA, Tripathy D, Gutheil JC, Harris LN, Fehrenbacher L, Slamon
DJ, Murphy M, Novotny WF, Burchmore M, Shak S, Stewart SJ, Press M. (2002)
Efficacy and safety of trastuzumab as a single agent in first-line treatment of
HER2-overexpressing metastatic breast cancer. J Clin Oncol. Feb 1;20(3):719-26.
von Boehmer H, Aifantis I, Gounari F, Azogui O, Haughn L, Apostolou I, Jaeckel E,
Grassi F, Klein L. (2003) Thymic selection revisited: how essential is it?
Immunol Rev. Feb;191:62-78. Review
von Boehmer H. (2005) Unique features of the pre-T-cell receptor alpha-chain: not just
a surrogate. Nat Rev Immunol. Jul;5(7):571-7. Review.
Vora A, Mitchell CD, Lennard L, Eden TO, Kinsey SE, Lilleyman J, Richards SM;
Medical Research Council; National Cancer Research Network Childhood Leukaemia
Working Party. (2006) Toxicity and efficacy of 6-thioguanine versus 6-mercaptopurine
in childhood lymphoblastic leukaemia: a randomised trial.
Lancet. Oct 14;368(9544):1339-48.
199
Walter AW, Hancock ML, Pui CH, Hudson MM, Ochs JS, Rivera GK, Pratt CB,
Boyett JM, Kun LE. (1998) Secondary brain tumors in children treated for acute
lymphoblastic leukemia at St Jude Children's Research Hospital.
J Clin Oncol. Dec;16(12):3761-7.
Wang J, Jani-Sait SN, Escalon EA, Carroll AJ, de Jong PJ, Kirsch IR, Aplan PD.
(2000) The t(14;21)(q11.2;q22) chromosomal translocation associated with T-cell acute
lymphoblastic leukemia activates the BHLHB1 gene.
Proc Natl Acad Sci U S A. Mar 28;97(7):3497-502.
Wang Y, Klijn JG, Zhang Y, Sieuwerts AM, Look MP, Yang F, Talantov D,
Timmermans M, Meijer-van Gelder ME, Yu J, Jatkoe T, Berns EM, Atkins D, Foekens
JA. (2005) Gene-expression profiles to predict distant metastasis of
lymph-node-negative primary breast cancer. Lancet. Feb 19-25;365(9460):671-9.
Warner JK, Wang JC, Hope KJ, Jin L, Dick JE. (2004) Concepts of human leukemic
development. Oncogene. Sep 20;23(43):7164-77. Review.
Weng AP, Nam Y, Wolfe MS, Pear WS, Griffin JD, Blacklow SC, Aster JC. (2003)
Growth suppression of pre-T acute lymphoblastic leukaemia cells by inhibition of
notch signalling. Mol Cell Biol. 23:655-664.
Weng AP, Ferrando AA, Lee W, Morris 4th JP, Silverman LB, Sanchez-Irizarry C,
Blacklow SC, Look AT & Aster JC. (2004) Activating mutations of NOTCH1 in
human T cell acute lymphoblastic leukaemia. Science. 306, 269–271.
200
Weng AP, Millholland JM, Yashiro-Ohtani Y, Arcangeli ML, Lau A, Wai C, Del
Bianco C, Rodriguez CG, Sai H, Tobias J, Li Y, Wolfe MS, Shachaf C, Felsher D,
Blacklow SC, Pear WS, Aster JC. (2006) c-Myc is an important direct target of Notch1
in T-cell acute lymphoblastic leukemia/lymphoma.
Genes Dev. Aug 1;20(15):2096-109.
Willemse MJ, Seriu T, Hettinger K, d'Aniello E, Hop WC, Panzer-Grümayer ER,
Biondi A, Schrappe M, Kamps WA, Masera G, Gadner H, Riehm H, Bartram CR, van
Dongen JJ. (2002) Detection of minimal residual disease identifies differences in
treatment response between T-ALL and precursor B-ALL.
Blood. Jun 15;99(12):4386-93.
Winter SS, Jiang Z, Khawaja HM, Griffin T, Devidas M, Asselin BL, Larson RS;
Children's Oncology Group. (2007) Identification of genomic classifiers that
distinguish induction failure in T-lineage acute lymphoblastic leukemia: a report from
the Children's Oncology Group. Blood. Sep 1;110(5):1429-38.
Wolfraim LA, Fernandez TM, Mamura M, Fuller WL, Kumar R, Cole DE, Byfield S,
Felici A, Flanders KC, Walz TM, Roberts AB, Aplan PD, Balis FM, Letterio JJ. (2004)
Loss of Smad3 in acute T-cell lymphoblastic leukemia.
N Engl J Med. Aug 5;351(6):552-9.
201
Wuchter C, Ruppert V, Schrappe M, Dörken B, Ludwig WD, Karawajew L. (2002)
In vitro susceptibility to dexamethasone- and doxorubicin-induced apoptotic cell death
in context of maturation stage, responsiveness to interleukin 7, and early cytoreduction
in vivo in childhood T-cell acute lymphoblastic leukemia.
Blood. Jun 1;99(11):4109-15.
Xia Y, Brown L, Yang CY, Tsan JT, Siciliano MJ, Espinosa R III, Le Beau MM, Baer
RJ. (1991) TAL2, a helix-loop-helix gene activated by the (7;9)(q34;q32) translocation
in human T-cell leukemia. Proc Natl Acad Sci U S A. Dec 15;88(24):11416-20.
Yan W, Young AZ, Soares VC, Kelley R, Benezra R, Zhuang Y. (1997) High
incidence of T-cell tumors in E2A-null mice and E2A/Id1 double-knockout mice.
Mol Cell Biol. Dec;17(12):7317-27.
Yeoh EJ, Ross ME, Shurtleff SA, Williams WK, Patel D, Mahfouz R, Behm FG,
Raimondi SC, Relling MV, Patel A, Cheng C, Campana D, Wilkins D, Zhou X, Li J,
Liu H, Pui CH, Evans WE, Naeve C, Wong L & Downing JR. (2002) Classification,
subtype discovery, and prediction of outcome in paediatric acute lymphoblastic
leukaemia by gene expression profiling. Cancer Cell. 1, 133–143.
Yuen T, Wurmbach E, Pfeffer RL, Ebersole BJ, Sealfon SC. (2002) Accuracy and
calibration of commercial oligonucleotide and custom cDNA microarrays.
Nucl Acids Res. 30:e48.
202
Yule SM, Price L, Cole M, Pearson AD, Boddy AV. (2001) Cyclophosphamide
metabolism in children following a 1-h and a 24-h infusion.
Cancer Chemother Pharmacol. 47:222-228.
Zhang H, Yu CY, Singer B. (2003) Cell and tumor classification using gene expression
data: construction of forests. Proc Natl Acad Sci U S A. Apr 1;100(7):4168-72.
Zhu YM, Zhao WL, Fu JF, Shi JY, Pan Q, Hu J, Gao XD, Chen B, Li JM, Xiong SM,
Gu LJ, Tang JY, Liang H, Jiang H, Xue YQ, Shen ZX, Chen Z, Chen SJ. (2006)
NOTCH1 mutations in T-cell acute lymphoblastic leukemia: prognostic significance
and implication in multifactorial leukemogenesis. Clin Cancer Res.
May 15;12(10):3043-9.
top related