Gene Profiling in Soft Tissue Sarcoma

#,#0-$'*',%', -$22'113# 0!-+ S 0#"'!2'4#4 *3#-$', 0!-+ 3+-30 0-%0#11'-, ," 304'4 *

Paromita (Romi) Das Gupta

A Thesis submitted to The University of New South Wales Faculty of Medicine for the Degree of Doctor of Philosophy (PhD)

August 2007

‘I hereby grant the University of New South Wales or its agents the right to archive and to make available my thesis or dissertation in whole or part in the University libraries in all forms of media, now or here after known, subject to the provisions of the Copyright Act 1968. I retain all proprietary rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part of this thesis or dissertation. I also authorise University Microfilms to use the 400 word abstract of my thesis in Dissertation Abstract International. I have either used no substantial portions of copyright material in my thesis or I have obtained permission to use copyright material; where permission has not been granted I have applied/will apply for a partial restriction of the digital copy of my thesis or dissertation.'

Signed ……………………………………………......

Date ……………………………………………...... AUTHENTICITY STATEMENT ‘I certify that the Library deposit digital copy is a direct equivalent of the final officially approved version of my thesis. No emendation of content has occurred and if there are any minor variations in formatting, they are the result of the conversion to digital format.’ '%,#"UUUUUUUUUUUUUUUUUTTTTTTTTTTTTTTTTTTTTTTTTTTT 2#UUUUUUUUUUUUUUUUUTTTTTTTTTTTTTTTTTTTTTTTTTTT

ORIGINALITY STATEMENT

The majority of the work described in this thesis was performed by myself, however important contributions were made by others. Cytogenetic analysis of the newly characterised cell lines was kindly performed by Dr Robyn Lukeis at SydPath, St Vincent’s Hospital, Sydney. TRAP assay for telomerase and APB staining for ALT in the above cell lines was performed by Jeremy Henson at Children’s Medical Research Institute, Sydney. KIT mutation analysis of the GIST-M cell line was carried out in collaboration with Maurice Loughrey and Victoria Beshay at the Peter MacCallum Cancer Centre, Melbourne. TP53 mutation analysis of the LMS-LFS cell line was carried out in collaboration with Dr Rodney Scott and his team at Hunter Area Pathology Service. Statistical analysis of gene expression data was carried out in collaboration with Dr Rohan Williams at the School of Biotechnology & Biomolecular Sciences (BABS), UNSW.

I hereby declare that this submission is my own work and to the best of my knowledge it contains no materials previously published or written by another person, or substantial proportions of material which have been accepted for the award of any other degree or diploma at UNSW or any other educational institution, except where due acknowledgement is made in the thesis. Any contribution made to the research by others, with whom I have worked at UNSW or elsewhere, is explicitly acknowledged in the thesis.

I also declare that the intellectual content of this thesis is the product of my own work, except to the extent that assistance from others in the project's design and conception or in style, presentation and linguistic expression is acknowledged.

Paromita (Romi) Das Gupta August 2007

ABSTRACT

Despite improvements in the clinical management of soft tissue sarcomas (STS), 50% of patients will die of metastatic disease that is largely unresponsive to conventional chemotherapeutic agents.

The aims of this study were to identify genes and pathways that are dysregulated in progressive and metastatic STS. In addition to this, cell lines from fresh tumours were initiated and established, thus increasing the repository of cell lines available for functional studies. Recent advances in the understanding of the molecular biology of STS have thus far not resulted in the use of molecular markers for clinical prognostication. Identifying novel genes and pathways will lead to molecular diagnostic methods to better stratify prognostic groups and could identify cellular targets for more efficacious treatments.

Gene expression profiling of sarcoma cell lines of increasing metastatic potential revealed over-expression of genes involved in the epidermal growth factor (EGF) and transforming growth factor beta (TGFβ) pathways. Factors involved in invasion and metastasis such as integrins and MMPs were over-expressed in the cell lines with higher metastatic potential. The developmental Notch pathway and cell cycle regulators were also dysregulated. NDRG1 was significantly over-expressed in the high grade sarcoma cell line, a novel finding in sarcomas. The expression of EGFR, NDRG1 and other genes from the above pathways was validated using quantitative RT-PCR in real time (qRT-PCR).

A tissue microarray (TMA) comprising STS of varying tumour grades was constructed for high throughput assessment of target proteins. EGFR, its activated form and its signal transducers were investigated using immunohistochemistry (IHC). Activated EGFR (HR 2.228, p < 0.001) and phosphorylated Akt (HR 2.032, p = 0.003) were found to be independent predictors of overall survival and both correlated with tumour grade.

Of the several STS cultures initiated and maintained, two of these cell lines were fully characterised in terms of cytogenetics, telomerase and alternate lengthening of

telomeres (ALT) status, KIT and TP53 mutation and the expression of certain biomarkers using both qRT-PCR and IHC.

In summary, transcript profiling identified several potential biomarkers of tumour progression and metastasis in STS. Crucially, activated EGFR and pAkt were found in a cohort of STS samples to correlate with clinical outcome, identifying them as potential diagnostic and therapeutic targets in the treatment of STS. Activated EGFR can be used as a diagnostic marker for patient selection, as well as for target effect monitoring. Furthermore, the cell lines established in this project will serve as valuable tools in future preclinical studies.

ACKNOWLEDGEMENTS

The translational laboratory based research that this PhD involved, required in a sense, a new beginning – the promise and challenge of being able to bring clinical and surgical training and knowledge to a different sphere of personal and mental growth. But as with most new challenges, one has to begin at the beginning and I am grateful to many people who were generous with their time and skills, for pointing me in the right direction.

I owe a debt of gratitude to my supervisor, Prof Phil Crowe, who has been a mentor and a sounding board through much of my surgical training as well as during the period while the Masters evolved into a PhD – because as always, there was more to do. It has been a tremendous support to have his encouragement, not just in the research realm but also at the coal face of personal and professional issues.

To my other supervisor Dr Jia Lin Yang, I owe many thanks. His perceptive help with the nuts and bolts of research strategy and enormous contribution to the statistical analysis of the patient data has been invaluable. Prof Ian Dawes, as another co- supervisor, was similarly insightful in his direction at the outset, when the research strategy was being formulated. Without his support and generosity, I would not have been able to carry out a substantial part of this research at the Ramaciotti Centre in UNSW.

I am indebted to Prof Pam Russell for taking me under her wing and over the years providing the right advice and encouragement. Her ability to perform rescue missions at precisely the right moments is a gift. I have greatly valued her enthusiasm and wisdom.

If one of the goals in embarking on a PhD is to acquire unofficial supervisors and turn them into lifelong friends, then I have executed this with aplomb. Dr Ruby Lin who happened to be at the Ramaciotti Centre, was co-opted into this role, which she willingly filled. As friend, co-conspirator and “supervisor”, her advice on matters intellectual, technical and otherwise, has been the source of much support over this time.

Many others at the Oncology Research Centre at Prince of Wales, at the Ramaciotti Centre, UNSW and other facilities have given of their time and provided much needed company in the midst of cell culture and RNA extraction. To Kim Ow, I owe many thanks, for his gentle instruction in immunostaining and his conversations on the more important things in life. Alex, Liz, Sheri, Hung, Lara and Michelle, who cheerfully answered my most basic questions at the start of this road. Ben, Bronwyn, Helen and Shining at the Ramaciotti Centre for company during those long microarray experiments. I am grateful to Dr Robyn Lukeis at St Vincent’s Hospital for her patience and guidance as I attempted to grasp the rudiments of cytogenetics. I also owe a debt of thanks to Drs Maurice Loughrey and Victoria Beshay at Peter Mac, Melbourne, Dr Jeremy Henson and Prof Reddel at CMRI, Westmead, Prof Rodney Scott at Hunter Area Pathology Services and Dr Louise Evans for their help in various aspects of this research.

On a more personal note, this work would simply not have been possible without my extended family and friends. The Wallace clan, from whom I have had much love and support. Ten tsi, my faithful hound and study buddy in the long hours of writing, Buddhist meditation dog, cannot go unmentioned. My husband, Mark, best friend and kindred spirit, there when all else is crashing. His unwavering love and belief in my ability has been an integral part of my sustenance.

Finally, and most importantly, everything I am and everything I strive to be, comes from my parents, Nikhil and Jharna. The best of childhoods was mine and ours was a house of laughter and love where I learnt what was most important and what cannot be taught.

Romi

&'15-0)'1"#"'! 2#"2-+7$ 2� &-*'4#"5'2&/3'#25'1"-+ ,"3, 0'"*#"#,2&31' 1+ ,"*#$25'2&!-30 %# ,"%0 !#3,2-3!&#"

PRESENTATIONS, AWARDS AND ABSTRACTS

JL Yang, R Das Gupta, IW Dawes, P J Crowe “Potential application of EGFR targeted therapies in different types of sarcomas” American Association for Cancer Research (AACR) 97th Annual Meeting, Washington D.C, April 2006.

R Das Gupta, RCY Lin, JL Yang, PJ Crowe “ Development of a Leiomyosarcoma cell line from a Li Fraumeni patient”, Surgical Research Society of Australasia 42nd Annual Scientific Meeting, in conjunction with Annual Scientific Congress, Royal Australasian College of Surgeons, Perth, May 2005. Winner of Young Investigator Award for best paper

R Das Gupta, RCY Lin, JL Yang, IW Dawes, PJ Crowe “Gene Expression profiling in Soft tissue Sarcomas”, 4th National Microarray Conference, Perth, Oct 2004. Winner of Travel Award from the Australasian Microarray and Associated Technologies Association (AMATA)

R Das Gupta, M Hannan, RCY Lin, IW Dawes, P J Crowe, JL Yang “Prognostic Factors in Soft Tissue Sarcoma using Gene Expression Profiling”, The Tow Prize Clinical and Research Day, Prince of Wales, Sydney Children’s and Royal Women’s Hospitals, Oct 2004

R Das Gupta, RCY Lin, JL Yang, IW Dawes, PJ Crowe “Development of a Tumour Progression Model in Soft Tissue Sarcoma using Gene Expression Profiling”, The Australian Health and Medical Research Congress, Darling Harbour, Sydney, Nov 2004

THESIS OVERVIEW

This thesis comprises eight chapters. The first chapter is a literature review, covering the salient aspects of the diagnosis and management of soft tissue sarcomas (STS), together with a more comprehensive overview of the molecular biology of STS. Current clinical prognostic markers and emerging molecular markers such as EGFR are discussed, with their relevance for targeted therapy. EGFR and other potential markers such as NDRG1 are explored in greater detail in subsequent chapters. Gene expression profiling as a technique for global gene and pathway discovery is introduced in the context of applying it to a STS tumour progression model.

The second chapter examines the global gene expression profiles of sarcoma cell lines of increasing metastatic potential with the aim of patterns of gene dysregulation associated with tumour progression and metastasis. The methods and results are presented and salient findings such as the differential expression of EGFR are discussed.

In the third chapter, the differential expression of EGFR and selected other candidate genes are validated using quantitative real time RT-PCR (qRT-PCR).

Chapter four introduces the Prince of Wales Hospital Sydney Sarcoma Clinic patient cohort. A tissue microarray is constructed from archival paraffin-embedded tumour specimens from this cohort in order to examine the protein expression of EGFR in clinical samples of varying histologic grade. EGFR, its activation and signal transduction is determined in this cohort using immunohistochemistry. The putative role of activated EGFR and phosphorylated Akt as diagnostic and prognostic markers in STS is discussed.

Chapters five through seven deal with the second concurrent aim of this study, which was to establish and characterise new sarcoma cell lines, to serve as in vitro models for future functional studies. In chapter five, the establishment of several short term cultures and the methods used for characterisation of two of the cell lines is described. Chapter six describes the characterisation of a new gastrointestinal stromal tumour cell line,

GIST-M. Chapter seven describes the characterisation of a cell line LMS-LFS, derived from a leiomyosarcoma in a patient with Li Fraumeni syndrome.

Chapter eight is a general discussion of the major findings in this study, in the context of their implications and future applications.

The appendices I to V contain detailed discussion of methodological issues in gene profiling, qRT-PCR and tissue microarray (TMA). Detailed protocols and additional data tables are also provided. Appendix II comprises lists of genes differentially expressed in the sarcoma cell lines of different metastatic potential.

Tables of contents

*#-$-,2#,21 ST ! " #" S &"' !() ," * "!+ ,TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTU STS ( )! (TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTV STT )# * 0-$ ! " #"TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTW STTTS ,!'"#,!# ,"-02 *'27TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTW STTTT '120' 32'-,TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT X STTTU '-*-%'! *#& 4'-30TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTX STU " "2 ( -$ S2") ("*"") #TTTTTTTTTTTTTTTY STUTS .-0 "'!TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTY STUTT " +'*' *7,"0-+#1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTY STV *" " ( ," " (-$ TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUS STVTS %'12-. 2&-*-%7TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUS STVTT &0 "',%TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUU STVTU 2 %',%TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUV STW ) "( -$ TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUY STX #"("#(-$ TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUZ STXTS 30%#07TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUZ STXTT ( "'-2� .7TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTU[ STXTU *&#+-2� .7TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVR STY ( " ', TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVS STYTS **','! *.0-%,-12'!+ 0)#01TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVS STYTT -*#!3* 0+ 0)#01TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVT STYTU -&"( 1 .0-%,-12'! 0)#0TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVX STZ & * 0-$ ! " #" TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVY STZTS *#**/',# %# '$$#0#,2' 2'-,TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVZ STZTT *72-%#,#2'!1S'+.*# ,"*-+.*#6TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTV[ STZTU 3#*-+#0# ',2#, ,!##!& ,'1+13TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTWZ ST[ # *!*"& * 0 ,"*+"($-0")2"0TTTTTTXR ST[TS "31'-,6,!-.0-2#',1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTXR ST[TT 370-1',#7', 1#1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTXS ST[TU 33+-3013..0#11-01TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTXT ST[TV 3',&' '2-01TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTXU ST[TW 8,%'-%#,#1'1',&' '2-01TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTXV ST[TX 3++ 07TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTX W STSR ,") !# ! (# )*', TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTXX STSRTS 33+-30&0-52& ,".0-%0#11'-,TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTXX STSRTT 3&##2 12 2'!.0-!#11TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTXZ STSRTU 3&#(-*#-$&#,#-6.0#11'-,.0-$'*',%TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTYR STSRTV : *'" 2'-,-$(#13*21',&#,#-6.0#11'-,23"'#1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTZW STSRTW 3 2#-$2&#!'#,!#TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTZX STSRTX 3'+1-$2&'13'1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTZY

Tables of contents

TT )( " ( -$ ( ," "2,"0 ', ! " #" ( 7(9 ( * (TTTTTTTTTTZZ TTS ( )! (TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTZ[ TTSTS #1'%,113#1',-6.0#11'-,.0-$'*',%TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTZ[ TTT #" "* "()#2 ) TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT[S TTTTS *#**/',#1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT[S TTTTT S[7800 71 ,";3 *'27*-,20-*-$.0',2',%TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT[U TTTTU -6.#0'+#,2 * #1'%,TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT[U TTTTV (<81-* 2'-,TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT[U TTTTW 811#11+#,2-$(<8;3 *'27TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT[V TTTTX ! <87,2'1Q/ #**',% ,"%7 0'"'1 2'-,TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT[X TTTTY ! ,,',%TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT[X TTTTZ 2 2'12'! *8, *71'1-$'!0- 00 7 2 TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT[Y TTU !* TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT[Z TTUTS *#**.0-*'$#0 2'-,( 2#1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT[Z TTUTT 811#11+#,2-$(<8;3 *'27 ,"/ #**#"! <8TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT[Z TTUTU ! ,,',%-$S[7+'!0- 00 71*'"#1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT[[ TTUTV (#13*21-$8, *71'1!-+. 0',%/-5 ,"%'%&%0 "#!#***',#1TTTTTTTTTTTTTTSRU TTUTW (#13*21-$8, *71'1!-+. 0',% **2&0##%0-3.1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSRZ TTV ) ! (TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSSV TTVTS 3&#>1#-$*#**/',#1$-0&#,#.0-$'*',%TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSSV TTVTT * ,"'" 2#&#,#1 ,"811-!' 2#". 2&5 71TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSSY TTVTU >8(?TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSTX UT +"* )" (-$"( 9 (-$ ," 2 "() )"& #"' TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTST[ UTS ( )! (TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSUR UTSTS 8"4 ,2 %#1-$(# *3'+#.*(TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSUS UTSTT <-0+ *'1 2'-,TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSUS UTSTU (#* 2'4#;3 ,2'2 2'-,TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSUS UTSTV * ,"'" 2#&#,#1,4#12'% 2#"TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSUT UTT #2 ) TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSUZ UTTTS *#***3*230#TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSUZ UTTTT (<8-620 !2'-, ,"! <87,2'1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSUZ UTTTU .0'+#0 #1'%,TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSU[ UTTTV (# *2'+#(3V.*(/(3.*(TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSU[ UTTTW 8% 0-1#&#*-*#!20-.&-0#1'1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSVS UTTTX (#* 2'4#/3 ,2'2 2'-,TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSVS UTU !* TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSVU UTUTS 8+.*'$'! 2'-,.*-21TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSVU UTUTT 811#11+#,2-$.*(.0-"3!2.#!'$'!'27TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSVU UTUTU 3&#&-31#)##.',%%#,#1%7&TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSVU UTUTV 8+.*'$'! 2'-,-$$'!'#,!'#1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSVV UTUTW (#* 2'4#/3 ,2'2 2'-,TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSWR UTV ) ! (TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSWW UTVTS * ,"'" 2#&#,#: *'" 2'-,TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSWW UTVTT *-,!-0" ,!#5'2&&#,#-6.0#11'-,800 71TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSW[ UTW 2" !##"0TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSXS

Tables of contents

VT ," ("*"( )! (', S *" (5'2& " ( ! #TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSXT VTS ( )! (TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSXU VTSTS 8"4 ,2 %#1-$2'113# 00 71TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSXV VTSTT : *'"'27-$38TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSXV VTSTU 8..*'! 2'-,-$3'113#'!0- 00 72#!&,-*-%7TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSXW VTSTV 3&#.0#1#,223"7TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSYR VTT #2 ) TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSYU VTTTS #*#!2'-,-$ 0!&'4 *1.#!'+#,1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSYU VTTTT **','! 2 1#TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSYU VTTTU *-,1203!2'-,-$38+ 12#0 *-!)1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSYV VTTTV 811#11+#,2-$',2#%0'27-$2'113#!-0#1-,38+ 12#0 *-!)1TTTTTTTTTTTTSYV VTTTW ++3,-&'12-!&#+'1207TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSYV VTTTX !-0',%-$++3,-12 ',',%TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSYW VTTTY **','!-. 2&-*-%'!!-00#* 2'-, ,"304'4 * , *71'1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSYW VTU !* TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSYZ VTUTS 33+-30%'12-*-%7 ,"&0 "#TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSYZ VTUTT ,2#%0'27-$2'113#!-0#1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSYZ VTUTU ,20 V33+-30: 0' '*'27 %#2#0-%#,#'27TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSY[ VTUTV -&"( ,"8!2'4 2#"-&"(TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSZS VTUTW .VV VT8.7-0)S -0)TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSZV VTUTX .8)2TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTS[R VTUTY .383UTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTS[S VTUTZ *-00#* 2'-, #25##,'-+ 0)#0-6.0#11'-, ,"33+-30&0 "#TTTTTTTTS[Y VTUT[ '-+ 0)#0-6.0#11'-, ,"%'12-*-%'!3 27.#TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTRR VTUTSR '1# 1#"0##304'4 *TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTRT VTUTSS64#0 **304'4 *TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTRW VTV ) ! (TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSV VTVTS *-00#* 2'-,5'2&304'4 *TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSV VTVTT '%,'$'! ,!#-$"',"',%1',2&'123"7TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSY VTVTU 3 0%#2#"3� .7TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTR VTW 3++ 07TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTU WT "&* 2#( ," #2 ) -$ 2"" " ( -$ (, " #"*** ( TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTV WTS ( )! (TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTW WTSTS *#***',#1 10#1-30!#1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTW WTSTT #2&-"1-$.0'+ 07*3*230#',/'2#0 230#TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTY WTSTU #2&-"1-$*& 0 !2#0'1 2'-,-$*#**/',#1',/'2#0 230#TTTTTTTTTTTTTTTTTTTTTTTTTTTTZ WTSTV .#!'$'!8'+1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT[ WTT #2 ) TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUR WTTTS *-**#!2'-,-$2'113#TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUR WTTTT .0'+ 07!3*230# ,"#12 *'1&+#,2-$!#***',#1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUR WTTTU *& 0 !2#0'1 2'-,-$!#***',#1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUR WTU !* TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUX WTUTS -12 *'1&+#,2-$.0'+ 07!3*230#1 ,"&-022#0+!#***',#1TTTTTTTTTTTTTTTTTUX WTUTT -0.&-*-%7-$&-023#0+*3*230#1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUY WTV ) ! (TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVV WTVTS .0- *#+15'2&+ ',2 ',',%!3*230#1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVW

Tables of contents

WTVTT #*#!2'-,-$!#***',#12-$302�!& 0 !2#0'1#TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVX XT & 0 !2#0'1 2'-, -$ ,-4#* V# !#** *',# $0-+ 120-',2#12', * 20-+ * 3+-30 0'1',% $0-+ 2&# )3-"#,3+TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVY XTS ( )! (TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVZ XTSTS **','! *.0#1#,2 2'-,TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVZ XTSTT %'12-%#,#1'1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVZ XTSTU 7332 2'-,1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTV[ XTSTV **','! *(#*#4 ,!#-$7332 2'-,1',&31TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTWR XTSTW *72-%#,#2'! #00 2'-,1 ,"&31TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTWR XTSTX &3!#***',#1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTWS XTT !* TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTWT XTTTS 33+-30-30!#TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTWT XTTTT -12 *'1&+#,2-$*#**/',#TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTWT XTTTU -0.&-*-%7TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTWT XTTTV ++3,-!72-!&#+'1207 ,"++3,-&'12-!&#+'1207TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTWW XTTTW *72-%#,#2'!1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTWX XTTTX 73 ,". &"(832 2'-,8, *71'1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTW[ XTTTY -6.0#11'-,-$'-+ 0)#01-,(# *3'+#(3V.*(TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTW[ XTTTZ 8, *71'1-$3#*-+#0# ',2#, ,!##!& ,'1+TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTXX XTU ) ! (TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTX[ XTUTS ++3,-.&#,-27.#TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTX[ XTUTT 73#6.0#11'-,TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTYR XTUTU 73+32 2'-, , *71'1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTYS XTUTV *72-%#,#2'!1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTYV XTUTW 3#*-+#0 1# ,"8/3TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTYW XTUTX -6.0#11'-,-$62�(#!#.2-0370-1',#7', 1#1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTYX XTUTY -6.0#11'-,-$73V 11-!' 2#"'-+ 0)#01TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTYY XTV !##"0 ," + TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTYZ YT 2"" " ( -$ ( +* *# V* ** * ( $0-+ * #0 " #"', . 2'#,25'2&* "!#( 0() #TY[ YTS ( )! (TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTZR YTT !* TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTZS YTTTS 33+-30-30!#TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTZS YTTTT -12 *'1&+#,2-$*#**/',#TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTZS YTTTU -0.&-*-%7TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTZT YTTTV ++3,-!72-!&#+'1207 ,"++3,-&'12-!&#+'1207TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTZT YTTTW *72-%#,#2'!1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTZZ YTTTX 3.WU32 2'-,8, *71'1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTZZ YTTTY 8, *71'1-$3S3(8.811 7 ,"8.12 ',',%TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT[U YTTTZ -6.0#11'-,-$'-+ 0)#01-,0# *V2'+#(3V.*(TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT[V YTU ) ! (TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT[X YTUTS /V/"!#***',#!-+. 0#"2-60'%', *33+-30TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT[X YTUTT -6.0#11'-,-$.(7*;-,0# *2'+#(3V.*(TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT[X YTUTU 3.WU+32 2'-,TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT[Y YTUTV *72-%#,#2'! #00 2'-,1',/V/"TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT[Y YTUTW 8/3 ,"2#*-+#0 1#',/V/"!#***',#TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT[[

Tables of contents

YTV !##"0 ," + TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTURS ZT ("* (*! ( ,"!! + TTTTTTTTTTTURT ZTS ( )! (TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTURU ZTT *- *#,#6.0#11'-, 22#0,1 ,"2&#3+-300-%0#11'-,#-"#* URV ZTTTS -&"(TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTURV ZTTTT 3&"TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTURV ZTTTU 62�* ,"'" 2#&#,#1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTURW ZTU "!2'4 2'-, ," '%, *0 ,1"3!2'-,', 0!-+ TTTTTTTTTTTTTTTTTTTTTTTTTTTTURY ZTV ,'2' 2'-, ,"& 0 !2#0'1 2'-,-$ 0!-+ #***',#1TTTTTTTTTTTTTTTTTTTTTTTTTTTURZ ZTW 3++ 07TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUR[ ZTX 3230#)'0#!2'-,1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUSR ZTXTS '-',$-0+ 2'!1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUSR ZTXTT "3,!2'-, *23"'#1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUSS ZTXTU >.120# +-$-&"(TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUSS ZTXTV 3 0%#2#"3� .7',3TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUST "()

T * ," ")) ("* )"" $-0 # ""0 ," **!*!9 #( TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUSW TS )#1'%, 113#1',6.0#11'-,0-$'*',%TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUSX TSTS (#.*'! 2#1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUSX TSTT -6.#0'+#,2 * #1'%,TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUSX TSTU .* 2$-0+S6*'%-,3!*#-2'"# 00 71TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUSY TSTV 7#' 1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUSZ TSTW <-0+ *'1 2'-, ,"8, *71'1-$ 2 TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUS[ TT #" "0 ** * " (TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUTT TU ?!"* 0 ( * ( ( * (!* )""0 TTTUTT TV (" *" (TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUTX TW !)(" 0(2 "()*"&** (TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUTY TX "("*0 # ""0 )"" ! ( & ()! bc " "*"'"TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUUS T ) ( "**09 )( Q ( * ," "2,"0 -,(9 ( * (-$ " #"** * ( TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUUU T ")) ("* * ," " "* "("*0 -$ +"* )" (-$(9 ()""31',%/TTTTTTTTTTTTTUXX TS 2'-, *#$-04 *'" 2'-,123"'#1',#,#0-$'*',%TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUXY TT ""4 ,2 %#1-$# *'+#TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUXZ TU (-0+ *'1 2'-,TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUXZ TV #* 2'4#?3 ,2'2 2'-,TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUX[ TVTS 3&#*-+. 0 2'4#*2"-0+3* TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUX[ TVTT (#* 2'4#;3 ,2'2 2'-,31',%(-3©TRRWTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUYR TW 2-31#)##.',% #$#0#,!##,#1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUYR TX ")) ("* !* TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUYU TXTS .*(-$$'!'#,!7TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUYU

Tables of contents

TXTT '$$#0#,2' *30 ,1!0'.2-6.0#11'-,TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUYZ +T ! # ""0S +"* ) 0Q * ," )" *))"""&* TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUY[ +TS + *'"'27-$'113##'!0- 00 7#"TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUZR :TSTS 33+-30#0-%#,#'27TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUZR :TSTT 8,2'%#,'!'27-$ 0!&'4 *1.#!'+#,1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUZS :TSTU ,2#0V ,"',20 V* -0 2-074 0' '*'27 ,"0#.0-"3!' '*'27TTTTTTTTTTTTTTTTTUZS :TSTV : *'" 2'-,',1-$22'113#23+-301TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUZT +TT "..*'! 2'-,-$2'113#+'!0- 00 72#!&,-*-%7TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUZT :TTTS 38',#.'2&#*' *+ *'%, ,!7TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUZT :TTTT 38',1-$22'113#1 0!-+ TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUZU +TU *'+'2 2'-,1-$'113##'!0- 00 7 ,"-3,2#0## 130#1TTTTTTTTTTTTTTTTTTTTTUZV :TUTS +.*#1'8#TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUZV :TUTT 33+-30#0-%#,#'27 ,"38 , *71'1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUZW :TUTU ,2#%0'27-$*-0#1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUZW :TUTV !-0',%-$++3,-12 ',',%TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUZX +TV *','! *" 2 . 2'#,21S 7",#7 0!-+ *','!) 2 1#TTTTTTTTTTTTTTTTTTTTUZZ +TW -,1203!2'-,-$#"+ 12#0 *-!)1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUZ[ :TWTS 3&#-/3'.+#,2TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUZ[ :TWTT #2&-"TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUZ[ :TWTU #1'%,-$2&# 00 71TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUZ[ :TWTV 811#11+#,2-$',2#%0'27-$2'113#!-0#1-,38 *-!)1TTTTTTTTTTTTTTTTTTTTTTTTTU[R +TX ++3,-12 ',',%0-2-!-*1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTU[R :TXTS 6.2'+'1 2'-,TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTU[R :TXTT ++3,-12 ',',%TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTU[S +TY 2'#,2& 0 !2#0'12'!1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTU[U +TZ 5) 2 $-0 ++3,-12 ',',%-$#"1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTU[V :TZTS -&"( ,"33+-30&0 "#TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTU[V :TZTT 8!2'4 2#"-&"( ,"33+-30&0 "#TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTU[V :TZTU .VV VT8.7 ,"33+-30&0 "#TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTU[V :TZTV .8)2 ,"33+-30&0 "#TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTU[W :TZTW .383U ,"33+-30&0 "#TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTU[W +T[ -00#* 2'-,-$)'1# 1#V0## 304'4 *5'2&*','! *+ 0' *#1TTTTTTTTTTTTTTTU[W +T * ," ")) ("* * ( "* )"" $-0 "&* 2#( ,"2"" " (-$*** ( TTTTTTTTTU[Y +TS -**#!2'-,-$2'113#TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTU[Z +TT 0'+ 07!3*230# ,"#12 *'1&+#,2-$!#***',#1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTU[Z +TU 0-2-!-*1$-0!& 0 !2#0'1 2'-,-$!#***',#1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTU[[ :TUTS -0.&-*-%7TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTU[[ :TUTT ++3,-&'12-!&#+'1207TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTU[[ :TUTU 32 2'-, *8, *71'1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVRS :TUTV 8, *71'1-$3#*-+#0# ',2#, ,!##!& ,'1+3TTTTTTTTTTTTTTTTTTTTTTTTVRT +TV 2'#,21 ," -30!#3+-301TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVRU :TVTS 6WTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVRU :TVTT 6XTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVRU :TVTU 6YTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVRV :TVTV 6ZTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVRV :TVTW 6[TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVRV

Tables of contents

:TVTX 6SRTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVRW :TVTY 6SSTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVRW :TVTZ 6STTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVRX :TVT[ 6SUTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVRX :TVTSR 6SVTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVRX :TVTSS 6SYTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVRY :TVTST 6SZTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVRY :TVTSU 6S[TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVRY :TVTSV 6TRTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVRZ :TVTSW 6TSTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVRZ :TVTSX 6TUTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVR[ :TVTSY 6TVTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVR[ :TVTSZ 6TWTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVSR +TW 2'#,2 ,"3+-30 -30!#$-0 V#) SW!#***',#TTTTTTTTTTTTTTTTTTTTTVSR :TWTS . 2'#,2.0#1#,2 2'-,TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVSR :TWTT %'12-*-%7-$33+-30TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVSS +TX 2'#,2 ,"3+-30 -30!#$-0*# V* ) SX!#***',#TTTTTTTTTTTTTTTTTTVST :TXTS . 2'#,2.0#1#,2 2'-, ,"%'12-*-%7-$33+-30TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVST :TXTT " +'*7.#"'%0##TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVST & &* "20TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVSV

*'12-$ *#1

3 *#STS%'12-*-%'!** 11'$'! 2'-,-$-$22'113#1 0!-+ 1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUW 3 *#STT%'12-*-%'!&0 "',%-$23'113# 0!-+ 1STTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUX 3 *#STU3+#0'! ,D-',2*-++'22##-,* ,!#08D**3<2 %',%-$-$2 3'113# 0!-+ 1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUX 3 *#STV**','!-. 2&-*-%'!.0-%,-12'!" !2-01$-0-620#+'27-$22'113#1 0!-+ VU 3 *#STW23"'#1-,-*#!3* 0.0-%,-12'!+ 0)#01',3QS[[XVTRRWTTTTTTTTTTTTTTTTTTTTTVW 3 *#STX"31'-,&#,#1', 0!-+ 1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTWT 3 *#STY*72-%#,#2'! #00 2'-,1',1-$22'113#1 0!-+ 1-$!-+.*#6) 07-27.#TTWY 3 *#STZ&#,##6.0#11'-,123"'#1-,! ,!#0 ,".0-%,-1'1',-.'2&#*' * *'%, ,!'#1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTYZ 3 *#ST[&#,#-6.0#11'-,23"'#1 ""0#11',%33+-30.0-%0#11'-,TTTTTTTTTTTTTTTTTTTTTTTTTTY[ 3 *#STSR&#,#-6.0#11'-,23"'#1-,-$22'113# 0!-+ 1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTZR 3 *#UTS.0'+#0131#"',0# *2'+#(3V.*(TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSVR 3 *#UTT2 2'12'! *8, *71'1-$ '$$#0#,2' *&#,#-6.0#11'-,TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSWU 3 *#VTS3'113#'!0- 00 7: *'" 2'-,23"'#1',62� *'%, ,!'#1TTTTTTTTTTTTTTTTTTTSXY 3 *#VTT3'113#'!0- 00 7123"'#1',62� *'%, ,!'#1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSXZ 3 *#VTU3'113#'!0- 00 7123"'#1', 0!-+ TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSX[ 3 *#VTV8,2' -"'#1#*#!2#"$-0++3,-&'12-!&#+'1207-,38TTTTTTTTTTTTTTTTTTTTTTTTTTTSYY 3 *#VTW,2#%0'27-$3'113#*-0#184 '* *#$-08, *71'1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSY[ 3 *#VTX,2#0, **-,20-*1811#11',%,20 V33+-30%#2#0-%#,#'27TTTTTTTTTTTTTTTTTTTTTTTTTSZR 3 *#VTY-&"(++3,-0# !2'4'27-$8**31 +.*#1 !0-11%'12-*-%'!&0 "#TTTTSZS

Tables of contents

3 *#VTZ*-00#* 2'-,-$-&"(#6.0#11'-,5'2&'%, *30 ,1"3!#01$-08** +.*#1 TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSZT 3 *#VT[8!2'4 2#"-&"(#6.0#11'-,', **31 +.*#1 !0-11%'12-*-%'!&0 "#SZU 3 *#VTSR*-00#* 2'-,-$8!2'4 2#"-&"(#6.0#11'-,5'2&'%, *30 ,1"3!#01',8** +.*#1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSZU 3 *#VTSS.VV VT8.7-0)S-0)T++3,-0# !2'4'27', **31 +.*#1TTTSZV 3 *#VTST*-00#* 2'-, #25##,-&"( ,".VV VT8.7-6.0#11'-,TTTTTTTTTTTTTTTTTTTTTTSZW 3 *#VTSU*-00#* 2'-, #25##,8!2'4 2#"-&"( ,".VV VT8.7-6.0#11'-,SZW 3 *#VTSV.&-1.&-07* 2#"8)2', **31 +.*#1 !0-11%'12-*-%'!&0 "#TTTTTTTTTTTS[R 3 *#VTSW*-00#* 2'-, #25##,-&"( ,".8)2-6.0#11'-,TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTS[S 3 *#VTSX*-00#* 2'-, #25##,8!2'4 2#"-&"( ,".8)2-6.0#11'-,TTTTTTTTTTTTTTTTTTTTS[S 3 *#VTSY.383U++3,-0# !2'4'27', **31 +.*#1 !0-1133+-30&0 "#TS[T 3 *#VTSZ*-00#* 2'-, #25##,-&"( ,".383U-6.0#11'-,TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTS[T 3 *#VTS[*-00#* 2'-, #25##,8!2'4 2#"-&"( ,".383U-6.0#11'-,TTTTTTTTTTTTS[T 3 *#VTTR7031) *F **'1S*-00#* 2'-, #25##,&0 "# ,"'-+ 0)#0-6.0#11'-, TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTS[Z 3 *#VTTS*&'V/3 0#2#121*-00#* 2',%'-+ 0)#0-6.0#11'-, ,"&0 "#TTTTTTTTTTTTS[[ 3 *# VTTT 7031) * F **'1 2#12 $-0 '-+ 0)#0 -6.0#11'-, ," %'12-*-%'! 3 27.# TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTRS 3 *#VTTU*-00#* 2'-,-$ '1# 1#"0##304'4 * ,"'-+ 0)#0-6.0#11'-,STTTRU VTTV 3*2'4 0' 2# *-6 (#%0#11'-, 8, *71'1 $-0 ,"#.#,"#,2 .0#"'!2-01 -$ 64#0 ** 304'4 *TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTRW 3 *#VTTW3*2'4 0' 2#8, *71'1$-064#0 **304'4 *-6!*3"',%<#- "(34 ,2* 1#1 TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTRX 3 *#VTTX>,'4 0' 2#8, *71'1$-064#0 **304'4 *TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTRX 3 *#WTST8,2' -"'#131#"',!& 0 !2#0'1 2'-,-$!#***',#1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUT 3 *#WTT.0'+#0131#"$-0+32 2'-, , *71'1-$&3V!#***',#TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUU 3 *#WTU.0'+#0131#"',0# *2'+#(3V.*($-0!& 0 !2#0'1 2'-,-$!#***',#1TTTTTTTTUX 3 *#WTV.0'+ 07*3*230#1','2' 2#"6!2- #0TRRUV 0!&TRRWTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUZ 3 *#XTS30 ,1!0'.2-6.0#11'-,',&3V ,"/V/"*#**/',#1TTTTTTTTTTTTTTTTTTTTTTTTTTXV 3 *#YTS++3,-12 ',',%0#13*21!-+. 0',%-0'%', *23+-30 ,"/V/"!#***',# -,!& + #01*'"#1 ," 1. 0 $$',#+ #""#".* 1+ !*-21TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTZV 3 *#TS%3+ ,S[7*-,20-*.-21TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUTV 3 *# TS &#,#.0',% , *71'1 !-+. 0',% /-5 %0 "# FXZV ," %'%& %0 "# %3SRZR!#***',#1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUUV 3 *#TT( , *71'1!-+. 0',%%'%& ,"/-5&0 "# 0!-+ *#**/',#1TTTTTTTTTTTTTTUUW 3 *# TU &#,#.0',% , *71'1 -$ "'$$#0#,2' **7 #6.0#11#" %#,#1 !0-11 2&# U 23+-30%0 "#1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUVR 3 *#TV%'#0 0!&'! ***312#0',%TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUVT 3 *#TW&#,#6,2-*-%'#1$-013 20##- 2 ',#"$0-+&'#0 0!&'! *!*312#0',%TTTTTTUVW 3 *#TX#2S$0-+)+# ,1!*312#0',%TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUVX 3 *#TY#2U$0-++# ,1!*312#0',%TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUWX 3 *# TZ ( , *71'1 -$ %#,#1 "'$$#0#,2' **7 #6.0#11#" !0-11 2&# 2&0## 23+-30 %0 "#1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUXS 3 *#TS '$$#0#,2' *-6.0#11'-,#25##,*#**/',#1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUYZ 3 *#:TS**','!-. 2&-*-%'!*& 0 !2#0'12'!1-$*-&-02TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTU[U 3 *#:TT-&"('++3,-12 ',',%',31 +.*#1 !0-11%'12-*-%'!&0 "#TTTTTTTTTTTTU[V 3 *#:TU8!2'4 2#"-&"(',31 +.*#1 !0-11%'12-*-%'!&0 "#TTTTTTTTTTTTTTTTTTTTTTTTTTTU[V

Tables of contents

3 *#:TV.VV VT8.7-0)S-0)T++3,-12 ',',%',31 +.*#1TTTTTTTTTTTTTU[V 3 *#:TW.&-1.&-07* 2#"8)2-6.0#11'-,',31 +.*#1 !0-11%'12-*-%'!&0 "# TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTU[W 3 *#:TX.&-1.&-07* 2#"383U',31 +.*#1 !0-1133+-30&0 "#TTTTTTTTTTTTTTTU[W 3 *#:TY*-00#* 2'-,-$ '1# 1#"0##304'4 * ,"**','! *: 0' *#1TTTTTTTTTTTTTTTTTTTTU[X

*'12-$'%30#1

"'%30#STS '120' 32'-,-$%'12-*-%'!13 27.#1-$3TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUT "'%30#STT7#7%#,#1',*',# %#"'$$#0#,2' 2'-,$0-+2&#.*30'.-2#,2+#1#,!&7+ * 12#+!#**TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVZ "'%30#STU30 ,1*-! 2'-,1',!72-%#,#2'! **71'+.*#1-$22'113#1 0!-+ 1TTTTTTTTTTTTTTTTTWU "'%30#STV*-+.*#633+-30V%-12,2#0 !2'-,1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTXX "'%30#STW3&##2 12 2'!* 1! "#TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTX[ "'%30#TTS-6.#0'+#,2 * #1'%,$-0&#,#.0-$'*',%-$ 0!-+ *#**/',#1TTTTTTTTTTTTTTTTTT[W "'%30#TTT37.'! *'-8, *71#08%'*#,20#13*231',%2&#(<8XRRR< ,-/ *&'. TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSRR "'%30#TTU811#11+#,2-$/ #**',%(# !2'-,TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSRS "'%30#TTV81! ,,#"'+ %#-$ &7 0'"'1#"S[71*'"#TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSRT "'%30#TTW&#,#6,2-*-%7$-0%#,#1-4#0#6.0#11#"',2&#%'%&%0 "#!#***',#TTTTTSRX "'%30#TTX:-*! ,-.*-21%#,#0 2#"5'2&'-!-,"3!2-0(. !) %#TTTTTTTTTTTTTTTTTTTTTTTTTTTSRX "'%30#TTY&#,#6,2-*-%7$-0"'$$#0#,2' **7#6.0#11#"%#,#1-,'-!-,"3!2-0(TSRY "'%30#TTZ%'#0 0!&'! *!*312#0',%-$%#,##6.0#11'-," 2 TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSSR "'%30#TT[&#,#-,2-*-%'#1-$%#,#1',!0# 1',%',#6.0#11'-,5'2&',!0# 1',% 23+-30%0 "#TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSSS "'%30#TTSR%'#0 0!&'! *!*312#0',%-$%#,##6.0#11'-," 2 TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSST "'%30#TTSS(#13*21-$)+# ,1!*312#0',%TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSSU "'%30#TTST '$$#0#,2' **7#6.0#11#"%#,#1',2&"J1'%, **',%. 2&5 7TTTTTTTTTTSTS "'%30#TTSU3&#8.71'%, **',%! 1! "#TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSTV "'%30#TTSV3&#',2#%0 2'-,-$1'%, **',%. 2&5 71TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSTY "'%30#UTS&#,#*-!31$-0-&"(TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSUU "'%30#UTT&#,#*-!31$-08!2'4',0#!#.2-037.#T8*:(TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSUU "'%30#UTU30 ,1$-0+',%%0-52&$ !2-0 #2 S 2S[/SUTSTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSUV "'%30#UTV&#,#*-!31$-0 ZZTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSUW "'%30#UTW8+.*'$'! 2'-,.*-21TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSVW "'%30#UTX.0-"3!2.#!'$'!'27TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSVX "'%30#UTY*-,$'0+ 2'-,-$.0-"3!2.#!'$'!'27TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSVY "'%30#UTZ.*( '*32'-,#0'#1 ,"#*2*304#1$-0-&"(TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSVZ "'%30#UT[.*(-$$'!'#,!7.*-21-$ -&"( ," 8.V7VTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSV[ "'%30#UTSR(#* 2'4#/3 ,2'2 2'-,-$%#,##6.0#11'-," 2 -,0# *V2'+#(3V.*(TSWV "'%30#UTSS'!0- 00 7 ,"/(3V.*(*-,!-0" ,!#TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSW[ "'%30#VTS-&"(1'%, **',%. 2&5 71TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSYT "'%30#VTT-&"(#6.0#11'-,TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSZX "'%30#VTU8!2'4 2#"-&"(#6.0#11'-,TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSZY "'%30#VTV-6.0#11'-,-$.VV VT8.7-0)S -0)TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSZZ "'%30#VTW.&-1.&-V8)2-6.0#11'-,TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSZ[ "'%30#VTX.&-1.&-V2 2U-6.0#11'-,TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTS[U

Tables of contents

"'%30#VTY-&"( ,"'%, *30 ,1"3!#0-*#!3*#-6.0#11'-,', (& "-+7-1 0!-+ TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTS[V "'%30#VTZ%-12 ',',%Q-&"( ,"'%, *30 ,1"3!#0-6.0#11'-,',8,%'-1 0!-+ TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTS[W "'%30#VT[8!2'4 2#"-&"( ,"'%, *30 ,1"3!#0-6.0#11'-,', .*#-+-0.&'! 0!-+ .TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTS[X "'%30#VTSR64#0 **304'4 *$-0. 2'#,2120 2'$'#" 7%'12-*-%'!&0 "#T%0-3.1 TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTR[ "'%30#VTSS64#0 **304'4 *$-0*-&-02120 2'$'#" 72 %#-$ '1# 1#TTTTTTTTTTTTTTTTTTTTTTTSR "'%30#VTST64#0 **304'4 *$-0. 2'#,2120 2'$'#" 733+-30'8#T%0-3.1TTTTSS "'%30#VTSU 64#0 **304'4 *-$. 2'#,2120 2'$'#"$-08!2'4 2#"-&"(-6.0#11'-, TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTST "'%30# VTSV 64#0 ** 304'4 * -$ . 2'#,21 20 2'$'#" $-0 .&-1.&-07* 2#" 8)2 .8)2 -6.0#11'-,TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTSU "'%30#WTS- 0*'#0.0'+ 07*3*230#1 11##,-,.& 1#!-,20 12+'!0-1!-.7TTTTTTTTTTTU[ "'%30#WTT-0.&-*-%7-$!#***',#"#0'4#"$0-+ ,3,"'$$#0#,2' 2#"1 0!-+ 6TUTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVR "'%30#WTU7,-4' *1 0!-+ ,".<-3*3*230#1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVT "'%30#WTV.0'+ 07!3*230#1 $2#0!-** %#, 1#"'%#12'-,TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVU "'%30#XTS.& 1#!-,20 12+'!0-1!-.7 ,"%-12 ',',%-$&3V!#***',#TTTTTTTTTTWU "'%30#XTT++3,-!72-!&#+'1207$-0&3V!#***',#TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTWV "'%30#XTU!73 ,":'+#,2',-6.0#11'-,TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTWW "'%30#XTV7 07-27.#-$&3V!#***',#TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTWZ "'%30#XTW+32 2'-, , *71'1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTXR "'%30#XTX-6.0#11'-,-$'-+ 0)#01$-0&3V ,"/V/"!#***',#1TTTTTTTTTTTTTTTTXT "'%30#XTY-6.0#11'-,-$.&-1.&-"'#12#0 1#T8 X ,"%'%&+- '*'27%0-3. -6S WTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTXU "'%30#XTZ-6.0#11'-,-$. &"(8 ,":-&"(TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTXW "'%30#XT[3#*-+#0 1#0#.# 2 +.*'$'! 2'-,.0-2-!-*3(8. 11 7TTTTTTTTTTTTTTTTTTXY "'%30#XTSR8*2#0, 2#*#,%2&#,',%-$2#*-+#0#18/3+#!& ,'1+TTTTTTTTTTTTTTTTTTTTTTTTTTTTXZ "'%30#XTSS203!230#-$230#!#.2-0 ,"*-! 2'-,-$+32 2'-,1TTTTTTTTTTTTTTTTTTTTTTTTTYU "'%30#YTS.#"'%0##-$/'"0 3+#,'17,"0-+#/"$ +'*7TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTZS "'%30#YTT60'%', *33+-30 ,"/V/"*#**/',#-0.&-*-%7TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTZU "'%30#YTU#*#!2#"0#13*21-$'++3,-12 ',',%$-0/V/"!#***',#-,!& + #0 1*'"#1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTZW "'%30#YTV#*#!2#"0#13*21-$'++3,-12 ',',%$-0/V/"!#***',#TTTTTTTTTTTTTTTTTTTTTTTTTZX "'%30#YTW++3,-12 ',',%$-07 'VS ,"7'2#,',',2&#/V/"!#***',#TTTTTTTTTTTTTTZX "'%30#YTX++3,-!72-!&#+'1207',/V/"!#***',#TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTZY "'%30#YTY*-+.-1'2#) 07-27.#-$2&#/V/"!#***',#TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT[R "'%30#YTZ*72-%#,#2'!8, *71'1-$/V/"!#***',#TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT[S "'%30#YT[!&#+ 2'!0#.0#1#,2 2'-,-$2&# [Y%#,#TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT[T "'%30#YTSR8/3#6.0#11'-,',/V/"!#***',#TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT[U "'%30#YTSS.(7*;#6.0#11'-,',/V/"!#***',#TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT[W "'%30#TS;3 *'27!-,20-*;*&7 0'"'1 2'-,1 28"#* '"#'!0- 00 7" !'*'27TTTTUTW "'%30#TS&#,#6,2-*-%7$0-+)+# ,1**312#0#2STTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUWW "'%30#TT3&#<-2!&. 2&5 7TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUXU "'%30#TU3&#,2#%0',. 2&5 7TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUXV "'%30#TV30 ,1$-0+',%%0-52&$ !2-0 #2 3&"$3,!2'-,1TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUXW

Tables of contents

"'%30#TS.*( '*32'-,#0'#1 ,"#*2*304#1$-0&8. %TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUYU "'%30#TT.*( '*32'-,#0'#1 ,"#*2*304#1$-0**< S ,"&8. %TTTTTTTTTTTTTTTUYV "'%30#TU.*(-$$'!'#,!7.*-21-$ 8*:(T ," &8. %TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUYW "'%30#TV.*(-$$'!'#,!7.*-21-$ 3&8T ," 3&8XTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUYX "'%30#TW.*(-$$'!'#,!7.*-21-$ < (&S ," **< STTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTUYY

Chapter 1: Soft tissue Sarcomas

CHAPTER ONE

1. SOFT TISSUE SARCOMA (STS): BACKGROUND and LITERATURE REVIEW

Chapter 1: Soft tissue Sarcomas

1.1 INTRODUCTION Soft tissue sarcomas are tumours of mesenchymal origin. The name is derived from the Greek words “sarcos” and “oma”, meaning fleshy tumour. They can arise wherever there is connective tissue, which includes smooth and skeletal muscle, blood vessels, adipose tissue and nerves. Sarcomas represent 1% of adult malignancies and a much higher proportion of childhood cancers (Enzinger and Weiss 1995b; Singer, Demetri et al. 2000; Grobmyer and Brennan 2003). The adult prevalence is approximately equivalent with that of carcinoma of the oesophagus, gliomas and multiple myeloma, and greater than that for Hodgkin’s disease. Disproportionate to their prevalence, sarcomas are responsible for 2% of cancer related deaths.

The central concept of this thesis is that gene dysregulation underpins the progression of the primary sarcoma to metastatic disease. The overall aim of this project was to identify patterns of gene dysregulation evident in the more aggressive tumours associated with poor clinical outcome, in order to drive patient selection for targeted treatment.

This chapter outlines what is known of the aetiology and pathogenesis of soft tissue sarcomas, together with the current treatment modalities of surgery, radiotherapy and conventional chemotherapy. The major focus, however, as with much of translational research in the field of oncology, is on the rapidly evolving field of gene discovery. The ultimate aim of such a research approach is the identification of biomarkers that may lead to the development of targeted therapies. Therapeutic agents currently in the early phases of pre clinical experimentation and/or clinical trial are producing mixed results, with evidence of emerging resistance. It is crucial, therefore, to further our understanding of the molecular pathogenic mechanisms involved in these tumours and to identify further targets that can be exploited for therapeutics.

Chapter 1: Soft tissue Sarcomas

1.2 EPIDEMIOLOGY of SOFT TISSUE SARCOMA

1.2.1 Incidence and Mortality

The incidence of soft tissue sarcomas has been difficult to estimate. The heterogeneous nature of soft tissue sarcomas and the controversies related to the classification of its various subtypes has contributed to this difficulty. In addition, early reports of tumours based on tumour location rather than histology has resulted in under-reporting of these mesenchymal malignancies (Sim, Edmonson et al. 1993; Storm 1994).

Data from the National Cancer Institute’s Surveillance, Epidemiology, and End Results Program (SEER) places the incidence of soft tissue sarcomas in the United States between 2.5 and 3 per 100000 (Pollock 2002). This translates to 8300 annual cases of soft tissue sarcomas (SEER 2002 estimate). Other sources indicate that there are approximately 4760 male and 3920 female new cases of STS annually (Cancer Facts and Figures, American Cancer Society, 2004). The incidence in other industrialised countries such as Australia and New Zealand are generally extrapolated from American statistics for STS. Soft tissue sarcomas affect a younger population of patients than epithelial tumours (70% in patients under 60 years of age). The incidence has risen over the last thirty years by up to 37% in certain age groups. Large increases of 16% (white males) and 146.5% (black males) have occurred particularly in young males aged under 19 (Pollock 2002). In Australia, there was an estimated 41% increase in incidence from 1988 to 1998. The Australian Institute of Health and Welfare predicts a 50% increase in incidence of bone and soft tissue sarcoma for women and a 32% increase for men between 2002 and 2011. While the rising rate of incidence may, in some part, be a reflection of the improved reporting and better diagnostic capability in recent years, the rise in incidence of this magnitude is significant.

The 5 year survival rate of all sarcomas is 50 – 60 % (Pisters 2002). Death from soft tissue sarcoma most often occurs from metastatic disease. 50% of patients presenting with soft tissue sarcomas will develop metastases but the figure is greater than 70% for those presenting with high grade tumours. 80% of metastases occur within two years of initial treatment (Pollock 2001). In the United States the number of sarcoma-related deaths was estimated to be 3660 for 2004 (Cancer Facts and Figures, American Cancer

Chapter 1: Soft tissue Sarcomas

Society, 2004). Due to the younger age of patients with soft tissue sarcoma, compared to patients with epithelial tumours, 17 life years per patient is lost with sarcoma related deaths (c.f. 6.5 years for breast, bowel and lung cancer). Like incidence, mortality rates have also risen over the last thirty years, despite advances in diagnostic and surgical technique.

1.2.2 Distribution

Soft tissue sarcomas, as noted above, can occur wherever mesenchymal tissue is present, in other words, anywhere in the body. Extremity sarcomas are most common (59%), followed by the trunk (19%), the retroperitoneum (15%) and the head and neck (9%) (Cormier and Pollock 2004).

1.2.3 Biological Behaviour

STS form a superficial or deep solid mass, with a centrifugal growth pattern and a surrounding pseudocapsule comprised of compressed tumour cells and reactive inflammatory tissue. The pseudocapsule may not be intact in higher grade malignancies, with tumour cells invading into and through it to cause skip lesions (Schmookler, Bickels et al. 2001). These are “metastases” that occur within the same anatomic compartment as the main tumour but are not considered distant metastases as they have not passed through the circulation. Skip lesions are thought to be involved in local recurrence. Sarcomas, unlike carcinomas, generally grow within the natural anatomic borders of the compartment but eventually invade adjacent tissue such as bone, vessels and nerves.

Despite their histologic heterogeneity, sarcomas follow a common metastatic pattern. The dominant pattern is haematogenous spread. Extremity sarcomas metastasise to the lungs, while retroperitoneal tumours spread to both liver and lungs. Lymph node metastases are rare (less than 5%), except for subtypes such as epithelioid sarcoma, rhabdomyosarcoma, clear-cell sarcoma, and angiosarcoma.

Chapter 1: Soft tissue Sarcomas

1.3 AETIOPATHOGENESIS of STS: THE TRADITIONAL PARADIGM

1.3.1 Sporadic

The aetiology, as with all malignancies, is multifactorial. There are a number of risk factors associated with the development of these tumours, including environmental carcinogens such as dioxin, herbicides such as phenoxyacetic acids and wood preservatives containing chlorophenols. Sarcomas have been known to arise in scar tissue after burns or surgical procedures or in limbs affected by chronic lymphoedema following axillary dissection (Stewart-Treves syndrome). External beam radiotherapy is a risk factor, with 0.1% of cancer patients treated with radiotherapy developing sarcomas in the irradiated field after a latency period of more than five years (Enzinger and Weiss 1995b). This represents an eight- to 50-fold increase in risk for patients treated with radiotherapy (Cormier and Pollock 2004). In these latter cases, the sarcomas are usually high grade malignant fibrous histiocytomas (MFH), with a poor survival rate. However, it has been shown that no secondary sarcomas developed in patients who had received <48Gy, while the absolute risk was 130/10,000 person-years for 160Gy (Kuttesch, Wexler et al. 1996). Most sarcomas have been reported to occur after exposure to 55Gy (Wei-wei, Qiu-liang et al. 2005). Other than the possible role of human herpes virus 8 (HHV-8) in HIV-1 associated Kaposi’s sarcoma, no other oncogenic virus is thought to play a role in the pathogenesis of sarcomas.

1.3.2 Familial Syndromes

Aside from the above risk factors associated with sporadic STS, a number of genetic syndromes are linked to a higher rate of STS development. The majority of these are the result of inherited germline mutations of tumour suppressor genes (detailed below), with the sarcomas and associated tumours arising with the “second hit” or loss of the wild type allele.

1.3.2.1 Li Fraumeni Syndrome

In 1969, Li and Fraumeni reviewed the records of 648 children with rhabdomyosarcomas (RMS) (Li and Fraumeni 1969a) and found four families with an overrepresentation of soft tissue sarcomas. The families were prospectively followed,

Chapter 1: Soft tissue Sarcomas

and it was revealed that other malignancies such as breast cancer occurred more frequently as well. The term “Li Fraumeni syndrome” (LFS) was later coined for the autosomal dominant condition and the diagnostic criteria defined (Pearson, Craft et al. 1982). Included within the spectrum of tumours were leukaemia, breast cancer, adrenocortical tumours, brain tumours and carcinoma of the pancreas (Garber, Goldstein et al. 1991). For those carrying the mutation, the probability of developing cancer is 50% by age 30, compared to only 1% for the general population in this age group. By age 70, 90% will have developed tumours. The relative risk of an individual with LFS developing a second cancer ranges from 1.5 to 83, depending on the age of onset of the first tumour (Hisada, Garber et al. 1998). The diagnosis of this inherited condition is therefore of great importance to patients and family members.

In 1990, germline TP53 mutations were confirmed as the underlying genetic basis for LFS (Malkin, Li et al. 1990; Srivastava, Zou et al. 1990). Since then, over 270 mutations of the TP53 gene have been reported in LFS, mostly involving the central DNA-binding region (Varley, Evans et al. 1997; Olivier, Eeles et al. 2002; Beroud and Soussi 2003). The mutant gene can exert a dominant-negative effect on the remaining wild type allele (Varley, Evans et al. 1997). It is significant that 30-60% of sporadic soft tissue sarcomas also have TP53 mutations (Hieken and Das Gupta 1996).

Mutations of the CHEK2 gene on chromosome 22q11, a putative tumour suppressor, have also been identified in a subset of patients with LFS who do not have detectable TP53 mutations (Bell, Varley et al. 1999), while mutations within the PTEN and CDKN2 genes have been excluded as possible causes (Burt, McGown et al. 1999).

In vitro studies of Li Fraumeni derived cell lines have additionally uncovered the use of the alternate mechanism for telomere maintenance (ALT) in spontaneously immortalised cells (Rogan, Bryan et al. 1995; Henson, Neumann et al. 2002; Tsutsui, Kumakura et al. 2003). This concurs with the finding that sarcomas, which are mesodermal in origin tend to be ALT positive, in contrast to epithelial tumours which employ telomerase as their telomere maintenance mechanism (TMM) (Shay and Bacchetti 1997; Colgin and Reddel 1999; Reddel 2003; Henson, Hannay et al. 2005). This is discussed in greater detail in the subsequent sections dealing with targeted

Chapter 1: Soft tissue Sarcomas

therapy (Section 1.9) and the characterisation of a new cell line derived from a soft tissue sarcoma in a patient with LFS (Chapter 5 Section 5.6).

The study of Li Fraumeni syndrome, with its attendant TP53 mutation and utilisation of ALT as the TMM, therefore provides invaluable insight into the genesis and progression of soft tissue tumours.

1.3.2.2 Retinoblastoma

A proportion of retinoblastomas are inherited. They tend to be bilateral, occur early in childhood and are the result of germline mutations of the RB1 gene on chromosome 13q14. Two thirds of second tumours that occur in affected individuals are mesenchymal in origin. Soft tissue sarcomas arising in patients with RB1 germline mutations include fibrosarcoma, leiomyosarcomas and liposarcoma (Friend, Horowitz et al. 1987) and deletions at the RB1 locus have also been found in sporadic cases of leiomyosarcoma, malignant fibrous histiocytoma, and undifferentiated sarcoma. In addition, soft tissue sarcomas expressing reduced levels of RB gene product have been shown to be more aggressive (Cance, Brennan et al. 1990).

The retinoblastoma gene is, like TP53, a recessive gene with “suppressor” function (Murphree and Benedict 1984). Tumours develop when both alleles of the gene are inactivated. Oncogenes, on the other hand, induce cancers following activation or alteration. RB1 can repress transcription of endogenous cell cycle genes, inhibiting progression from G1 to S phase of the cell cycle. This is achieved by complexing with the E2F family of transcription factors. Repression by the RB-E2F complex mediates the G1 arrest triggered by transforming growth factor-beta (TGFB) and p16(INK4A) or cyclin-dependent kinase inhibitor 2A (CDKN2A) (Zhang, Postigo et al. 1999).

1.3.2.3 Neurofibromatosis Type 1

Neurofibromatosis type I (NF1) is an autosomal dominant disorder caused by mutation in the neurofibromin gene on chromosome 17q11.2. It manifests as cafe-au-lait spots and neurofibromas of the skin. As early as 1956, it was noted that this disorder is associated with a higher than usual incidence of malignant neoplasms (Crowe, Schull et al. 1956). In an extensive review of 678 patients with neurofibromatosis, 21 malignant

Chapter 1: Soft tissue Sarcomas

neoplasms were found (D'Agostino, Soule et al. 1963). Secondary tumours include fibrosarcomas, neurofibrosarcomas and rhabdomyosarcomas (Knight, Murphy et al. 1973; Crawford 1986). The most common associated soft tissue sarcoma is the malignant peripheral nerve sheath tumour (MPNST), the lifetime risk for which in individuals with NF1 is 8%-13% (Evans, Baser et al. 2002). The median age at diagnosis of MPNST in NF1 patients in this study by Evans et al was 26 years, compared to 62 years in patients with sporadic MPNST.

Over 80% of germline mutations appear to result in truncation of the neurofibromin protein and consequent activation of the RAS pathway. No evidence of loss of heterozygosity has been observed in neurofibromas and it has been postulated that mutation in a second gene is required for tumorigenesis (Shen, Harper et al. 1996).

1.3.2.4 Other Genetic Syndromes

In a review of ten sarcoma-prone families, Lynch et al showed an association with sarcomas in families with germline mutations of BRCA1, MSH2, CDKN2A (Lynch, Deters et al. 2003). Their interest was sparked by their own earlier report of familial atypical multiple-mole melanoma-pancreatic carcinoma syndrome (FAMMM-PC) (Lynch, Brand et al. 2002). Eight families were confirmed to have germline CDKN2A mutations but their phenotypic characteristics varied, with melanoma predominating in some families and pancreatic carcinoma in others. One of these eight families had two family members with sarcoma.

The more recent study focused on families with hereditary syndromes, where, apart from one family with Li Fraumeni syndrome, sarcomas were not a key phenotypic feature. Two families known to have hereditary non-polyposis coli (HNPCC) with a germline mutation in the DNA mismatch-repair gene MSH2 had family members with both osteosarcoma and soft tissue sarcoma. Another family with BRCA1 mutations associated primarily with breast and ovarian cancer had a member with chondrosarcoma at the age of twenty seven.

The common thread in the above syndromes is that as with TP53, RB1 and NF1, they all involve inactivating mutations of tumour-suppressor genes, inherited in autosomal

Chapter 1: Soft tissue Sarcomas

dominant fashion. They are integral to maintaining genomic integrity and their disruption results in genomic instability. BRCA1 and MSH2 are involved in DNA- damage repair (Scully and Livingston 2000; Bernstein, Bernstein et al. 2002). When damaged DNA is not repaired, apoptosis of the cell is triggered via TP53 related mechanisms. If TP53 is non functional, cells with damaged DNA continue to proliferate, potentially leading to tumorigenesis (Schuyer and Berns 1999; Bernstein, Bernstein et al. 2002).

In familial gastrointestinal stromal tumours (GISTs), the gain of function mutations in KIT differ from the other syndromes above. These mutations result in constitutive activation of KIT, resulting in phosphorylation and activation of a variety of downstream signal transduction targets (Heinrich, Rubin et al. 2002). GISTs and the associated KIT mutations are discussed in greater detail in the subsequent sections of this thesis dealing with targeted therapy (Section 1.9) and the characterisation of a new cell line derived from a malignant gastrointestinal stromal tumour (Chapter Five, Section 5.5).

1.4 CLASSIFICATION and STAGING of STS

1.4.1 Histopathology

There are over 50 histological subtypes of STS described in the literature. They are typically named according to the predominant cell type within the tumour and its resemblance to the corresponding normal tissue or, in some cases, its embryonal counterpart (Enzinger and Weiss 1995b). Some of the more common subtypes are presented in Table 1.1.

The distribution of the various histological subtypes varies with tumour location (extremity versus retroperitoneal). The histologic distribution for adult tumours is classically thought to be as shown in Figure 1.1. Estimating the relative frequencies of the major histological types has been confounded by the reclassification of tumours. In 2002, the World Health Organisation (WHO) amended and expanded the classification of soft tissue tumours (Hogendoorn, Collin et al. 2004). Among the changes included

Chapter 1: Soft tissue Sarcomas

recognition of four different types of liposarcoma (LPS): well differentiated, dedifferentiated, myxoid and pleomorphic.

The debate over the concept of malignant fibrous histiocytomas (MFH) was acknowledged, in relation to its cell of origin. This was originally thought to be the histiocyte, then a “facultative” fibroblast and now either a fibroblast (Hsu, Rohol et al. 1989; Antonescu, Erlandson et al. 2000; Suh, Ordonez et al. 2000; Nakatani, Marui et al. 2001; Erlandson and Antonescu 2004) or a pluripotent mesenchymal cell. It was proposed that all MFH should be reclassified according to the predominant line of differentiation or phenotype exhibited, namely pleomorphic fibrosarcoma, myxofibrosarcoma, giant cell fibrosarcoma and lastly, pleomorphic sarcoma, not otherwise specified (NOS) (Erlandson and Antonescu 2004).

Soft issue Sarcoma Histology

Kaposi SS 4% MPNST 5% MFH 26% 9%

RMS 9%

FS 11% LPS LMS 24% 12%

Figure 1.1 Distribution of Histologic subtypes of STS. The most commonly occurring sarcomas are Malignant fibrous histiocytoma (MFH) and Liposarcoma (LPS). FS: Fibrosarcoma, LMS: Leiomyosarcoma, MPNST: Malignant peripheral nerve sheath tumour, RMS: Rhabdomyosarcoma, SS: Synovial sarcoma

Pleomorphic MFH has become synonymous with undifferentiated high grade sarcoma (Hogendoorn, Collin et al. 2004), representing perhaps a final common pathway of dedifferentiation for all sarcomas (Brooks 1986). The pleomorphic sarcomas of myogenic origin (leiomyosarcomas, rhabdomyosarcomas), have a worse prognosis than liposarcomas (Fletcher, Gustafson et al. 2001; Slavin and Thomas 2006). Given that

Chapter 1: Soft tissue Sarcomas

MFH and LPS are the two most commonly occurring types of STS, these changes will have considerable impact on prognostic data.

1.4.2 Grading

In addition to histologic type, such information as grade and stage are important to assess prognosis. The original grading system proposed by Broders and colleagues in 1939 (Enzinger and Weiss 1995b; Pollock 2002) was based on cellularity, pleomorphism, mitotic activity, necrosis, and type of growth (expansive versus invasive). Since then, three and four grade systems have been proposed; the two most common systems in current use being the four grade system of the National Cancer Insititute (NCI) (Markhede, Angervall et al. 1982; Costa, Wesley et al. 1984) and the three grade French Federation of Cancer Centers (FNCLCC) system (Coindre, Trojani et al. 1986; Coindre, Nguyen et al. 1988; Coindre 1993). The latter system utilises the histologic criteria of mitotic rate, degree of tumour differentiation and nuclear pleomorphism to assign numerical values, from which a composite number of 1-3 is generated, as detailed in Table 1.2.

Grading of tumours can traditionally only be applied to untreated primary tumours. Resected tumours that have had neoadjuvant chemoradiotherapy are excluded, as this may have altered the degree of necrosis and mitotic rate. Nor can the grading system be applied to types of tumours that do not have an identifiable normal cell counterpart, such as alveolar rhabdomyosarcoma. Alveolar and embryonal rhabdomyosarcomas are considered to be high grade sarcomas, as are neuroblastomas, extraskeletal Ewing’s sarcomas and peripheral neuroepitheliomas (Enzinger and Weiss 1995a). Gastrointestinal stromal tumours (GISTs) are currently graded based on factors such as tumour size, location and mitotic rate and hence fall outside the FNCLCC system.

Paediatric tumours are also excluded from the FNCLCC system, although they are usually assigned a grade by pathologists. Numerous studies have examined clinical prognostic factors in STS, finding histologic grade or features thereof, such as mitotic rate, to be the most significant independent prognostic variables for predicting local recurrence, metastasis and survival (Markhede, Angervall et al. 1982; Myhre-Jensen,

Chapter 1: Soft tissue Sarcomas

Kaae et al. 1983; Tsujimoto, Aozasa et al. 1988; Coindre, Terrier et al. 1996; Coindre, Terrier et al. 2001; Trassard, Le Doussal et al. 2001; Kaytan, Yaman et al. 2003; Koea, Leung et al. 2003; Zagars, Ballo et al. 2003a; Zagars, Ballo et al. 2003b; Svarvar, Bohling et al. 2007; ten Heuvel, Hoekstra et al. 2007).

Over the last twenty five years, histologic tumour grade, out of all clinical prognostic indicators, has consistently been shown to be the strongest predictor of metastasis and mortality (Zagars, Ballo et al. 2003a; Zagars, Ballo et al. 2003b). Investigating the genetic factors associated with tumour grade may therefore hold the key to understanding tumour progression and metastasis in soft tissue sarcomas.

1.4.3 Staging

The American Joint Committee on Cancer (AJCC) system for STS is based on the Tumour-Node-Metastasis (TNM) staging system, with the incorporation of a four tier histologic grading score (Fleming, Phillips et al. 1996; Fleming 2001). Staging incorporates the most useful prognostic indicators of size, depth and grade, as determined by the above studies. The current challenge remains to identify the molecular markers that distinguish the high risk patients, to establish a “molecular staging” model that will facilitate the application of targeted therapies for these patients.

Chapter 1: Soft tissue Sarcomas

Table 1.1 Histologic Classification of Soft tissue sarcomas .0#"-+', ,2* 2#%-07 3 27.#1 "' 0-31"' 0-1 0!-+ *-,%#,'2 *$' 0-1 0!-+ ,$* ++ 2-07$' 0-1 0!-+ "' 0-&'12'-!72'! #0+ 2-$'"' 0-1 0!-+ .0-203 #0 ,1 ". *'%, ,2$' 0-31&'12'-!72-+ "% 2-0'$-0+V.*#-+-0.&'! 76-'"$' 0-31 &' ,2!#** ,$* ++ 2-07 H ,2&-+ 2-31 /'.-+ 2-31/'.-1 0!-+ /. F#**"'$$#0#,2' 2#" 76-'" (-3,"!#** .*#-+-0.&'! #"'$$#0#,2' 2#" +--2&+31!*#/#'-+7-1 0!-+ / -.'2&#*'-'"*#'-+7-1 0!-+ )#*#2 *+31!*#(& "-+7-1 0!-+ ( -+ 07-, * 8*4#-* 0 -207-'" .',"*#!#** .*#-+-0.&'! *--"4#11#*1 /7+.& 2'!1 8,%'-1 0!-+ /7+.& ,%'-1 0!-+ 7 .-1'1 0!-+ .#0'4 1!3* 0 *'%, ,2& #+ ,%'-.#0'!72-+ 7,-4' *7,-4' *1 0!-+ -,-.& 1'!"'0-31-0-.'2&#*' * '.& 1'!"'0-31 ,"-.'2&#*' * <#30 * *'%, ,2.#0'.� *,#04#1&# 2&23+-301 .<3 **# 0!#**1-$22'113#1 0!-+ **3# *'%, ,2%0 ,3* 0!#**23+-30 .0'+'2'4#,#30-#!2-"#0+ *23+-30.<-3 <#30- * 12-+ < & ,%*'-,#30- * 12-+ -620 1)#*#2 *-5',%_11 0!-+ -620 1)#*#2 *! 02'* %',-31 -11#-31 -620 1)#*#2 * !&-,"0-1 0!-+ -12#-1 0!-+ >,!#02 ',&'12-%#,#1'1 8*4#-* 01-$2. 021 0!-+ -.'2&#*'-'"1 0!-+ This is not presented as an exhaustive list of all subtypes of soft tissue sarcoma, nor are the designated “predominant categories” always indicative of histogenesis. *CCSST is also known as Malignant melanoma soft parts (MMSP).

Chapter 1: Soft tissue Sarcomas

Table 1.2 Histologic Grading Soft Tissue Sarcomas: French Federation of Cancer Centers (FNCLCC) system . 0 +#2#0 "',"',% .-',21 "#%0##-$23+-30 !*-1#0#1#+ * ,!#2-,-0+ * "3*22'113#5#**V S "'$$#0#,2' 2'-, "'$$#0#,2' 2#" 23+-3027.#!*# 0*70#!-%,'8 *# T 23+-3027.#3,!#02 ',.--0*7-03,"'$$#0#,2' 2#" U 23+-30,#!0-1'1 ,-23+-30,#!0-1'1-, ,71*'"# R 23+-30,#!0-1'10WR$ S 23+-30,#!0-1'1≥WR$ T +'2-2'!!-3,2 RV[.#0SR&'%&.-5#0#"$'#*"1 S SRVS[.#0SR&'%&.-5#0#"$'#*"1 T ≥TR.#0SR&'%&.-5#0#"$'#*"1 U Total Score = (tumour differentiation) + (tumour necrosis) + (mitotic count) !-0# %'12-*-%'!&0 "# T-0U &0 "#S V-0W &0 "#T XY-0Z &0 "#U

Table 1.3 The American Joint Committee on Cancer (AJCC) TNM Staging of Soft Tissue Sarcomas TNM Criteria . 0 +#2#0 "',"',% #1'%, 2'-, .0'+ 0723+-30 23+-30≤W!+ 3S 23+-301W!+ 3T "#.2& 13.#0$'!' * 3!-"# "##. 3!-"# 0#%'-, **7+.& ,-&'12-*-%'! **74#0'$'#"+#2 12 1'12-

Chapter 1: Soft tissue Sarcomas

Staging #1!0'.2'-, & 3< 2 %#8 *-5%0 "#1+ **13.#0$'!' * ," &S&T 3S

1.5 DIAGNOSIS of STS

Soft tissue sarcomas typically present as a mass, with or without associated symptoms such as pain or loss of function as a consequence of the mass. Retroperitoneal tumours may manifest clinically when they displace, invade and/or obstruct major organs. Diagnosis requires the combined modalities of imaging and biopsy. For extremity lesions, magnetic resonance imaging (MRI) provides invaluable prognostic and surgical information regarding involvement of contiguous structures and neurovascular bundles. Abdominal, pelvic or mediastinal tumours are evaluated with computed tomography (CT). Core biopsies, incision or excision biopsies are required for histologic diagnosis. The principle to be adhered to in obtaining specimens is to sample sufficient tissue for diagnosis while causing minimal disruption of adjacent structures or the tumour itself. A core or incisional biopsy allows sufficient tissue to be obtained for cytogenetic analysis, electron microscopy and flow cytometry, in addition to histology and grading (Hunt, Vorburger et al. 2002).

An incisional biopsy should ideally be performed at a specialist centre, by the surgeon who will perform the definitive surgery. The incision should be oriented such that the subsequent wide local excision can incorporate the biopsy site and the scar (Shiu and Brennan 1989; Yasko 2002; Cormier and Pollock 2004). A poorly planned incisional biopsy may lead to an excessively large defect at wide local excision or necessitate a

Chapter 1: Soft tissue Sarcomas

wider field of postoperative irradiation. In the case of abdominal, pelvic or mediastinal sarcomas, CT-guided core biopsies can be carried out, applying the same principles outlined above.

1.6 MANAGEMENT of STS

1.6.1 Surgery

Surgery is the cornerstone of management for soft tissue sarcomas. The goal of surgical intervention is to achieve clear excision margins and preserve or restore function, wherever possible. The biopsy incision or core biopsy needle tract is included en bloc with the resected specimen to prevent local recurrence along the tract. For extremity sarcomas, wide local excision or limb preserving surgery has largely replaced compartmental excision and amputation, with adequately performed surgery and radiotherapy achieving equivalent local recurrence rates (Wilson, Davis et al. 1994; Yang, Chang et al. 1998). Enucleation of the tumour with its surrounding pseudocapsule is avoided as malignant cells are known to extend through the pseudocapsule. Resection of involved vascular structures followed by reconstruction can be combined with this limb- sparing approach. The importance of achieving clear margins has been proven in a number of large studies, all of which demonstrated a higher local recurrence rate with positive resection margins or margins less than 1 cm (Pisters, Leung et al. 1996; Baldini, Goldberg et al. 1999; Stojadinovic, Leung et al. 2002a). Margins may however be more difficult to determine in the case of retroperitoneal sarcomas (Stojadinovic, Leung et al. 2002a). Positive margins, in some of these studies were also found to predict disease free survival. The evidence for the association with the occurrence of metastases has been more contradictory. Positive resection margins did not predict the development of metastases in two studies involving 559 and 1041 patients (Pisters, Leung et al. 1996; Trovik, Bauer et al. 2000). Adjuvant or neoadjuvant radiotherapy is required, however, to achieve acceptable rates of local control (see section below).

Surgery is also the treatment of choice for isolated metastatic disease (Putnam Jnr 2002). The criteria for patient selection for surgery in the case of pulmonary metastases must include absence of uncontrolled extrathoracic disease, the potential for complete

Chapter 1: Soft tissue Sarcomas

resection and sufficient pulmonary reserve following resection for there to be a survival benefit. Reoperations for recurrent pulmonary metastases can also be performed for complete resection, achieving improved survival. Surgery can be combined with preoperative chemotherapy, using doxorubicin, cyclophosphamide and dacarbazine. The effect of multimodal treatment in these cases on post-thoracotomy survival is variable and cannot be predicted by preoperative response to the chemotherapy (Pollock 2002).

1.6.2 Radiotherapy

1.6.2.1 External Beam and Brachytherapy

A number of randomised trials, prospective and retrospective studies have evaluated the benefits of radiotherapy (Yang, Chang et al. 1998; Komdeur, Hoekstra et al. 2002; Borden, Baker et al. 2003; Strander, Turesson et al. 2003; Cormier and Pollock 2004; Scoggins and Pollock 2005). Combined surgery and external beam radiotherapy has been unequivocally shown to deliver better local control rates than surgery alone, irrespective of tumour grade (Yang, Chang et al. 1998). Similar results have been attained combining brachytherapy with surgery for high grade tumours (Strander, Turesson et al. 2003; Laskar, Bahl et al. 2007). The advantage of the latter technique is that it allows a higher dose of radiotherapy to be delivered directly to the tumour bed over a shorter period of time. Radiotherapy has not, however, improved survival.

1.6.2.2 Neoadjuvant versus Postoperative Radiotherapy

Proponents of neoadjuvant radiotherapy assert that lower doses of radiation can be delivered over a smaller field to an undisturbed tumour bed. In addition, preoperative radiotherapy may shrink the tumour, facilitating surgical resection. A randomised control trial comparing preoperative and postoperative radiotherapy for extremity sarcomas showed identical local control rates (93%) and equivalent disease free survival in the two groups (O'Sullivan, Davis et al. 2002). Wound infections were, however, more common in the neoadjuvant group.

1.6.2.3 Neoadjuvant Chemoradiotherapy

Over the last 20 years, a combination of chemotherapy and radiotherapy in the neoadjuvant setting has been successfully applied to deep extremity soft tissue sarcomas with the aim of improving local recurrence rates (Eilber, Morton et al. 1984; Pisters,

Chapter 1: Soft tissue Sarcomas

Ballo et al. 2002). The chemotherapeutic agent in this instance serves to sensitise the tumour to radiotherapy. Other variations of the neoadjuvant protocol instituted by Eilber and colleagues have been reported, such as in a multicentre series of 55 patients (Wanebo, Temple et al. 1995) and a cohort of 42 patients at the Tom Baker Cancer Centre in Canada (Temple, Temple et al. 1997). The cohort from the Tom Baker Cancer Centre has since increased to 75 patients with a median follow up of 6.75 years (Mack, Crowe et al. 2005). The limb salvage and local control rates of 95% and 94% at 5 years were similar to the previous studies. An additional advantage was the acceptable wound complication rate that was reported in this study.

1.6.3 Chemotherapy

1.6.3.1 Isolated limb perfusion

Isolated limb perfusion (ILP) is one technique that can be applied to advanced, multifocal or recurrent extremity sarcoma. Larger doses can be delivered directly to the affected limb, once “isolation” of the limb is secured by clamping the major artery and vein, ligating the collaterals and connecting the limb to an extracorporeal circuit. ILP with Doxorubicin, combined with preoperative external beam radiotherapy has been employed for extremity sarcomas that would otherwise have required amputation (Hegazy, Kotb et al. 2007). A limb salvage rate of 82.5% was achieved in this study. Melphalan and tumour necrosis factor-alpha (TNFα) have been used in combination or as single agents. The results of various trials confirmed > 80% limb salvage rates1 (Rossi, Mocellin et al. 2003; Strander, Turesson et al. 2003; Grunhagen, Brunstein et al. 2004).

1 Lower limb salvage rates are associated with isolated limb perfusion carried out for recurrent sarcomas, including sarcomas that recurred within the irradiated field 40

Chapter 1: Soft tissue Sarcomas

1.6.3.2 Systemic chemotherapy

Doxorubicin and ifosfamide are the most common agents in the adjuvant treatment of STS. Dacarbazine has been used in combination with the above. The benefits of chemotherapy for STS are still debated. The meta-analysis carried out for the Cochrane database ((SMAC) 1997) provided evidence that doxorubicin-based chemotherapy improved the time to local and distant recurrence. There was a trend towards improved overall survival but it was not statistically significant. A smaller Italian trial did show improved survival with epirubicin and ifosfamide (Frustaci, Gherlinzoni et al. 2001). No survival advantage has been documented when chemotherapy is used in the setting of retroperitoneal sarcomas, unresectable or metastatic disease.

1.7 PROGNOSTIC FACTORS in STS

1.7.1 Clinical prognostic markers

The clinical and histologic parameters used for prognostication have been discussed briefly in the section on histologic grading of STS (Section 1.4.2). A review of 119 patients with extremity soft tissue sarcomas treated at our hospital in Prince of Wales, Sydney, between 1972 and 1992 found recurrent presentation, patient age, tumour size and histologic grade to be independent predictors of survival (Wilson, Crowe et al. 1999).

The independent prognostic factors as determined by regression analysis in a large study of 1041 patients are shown in Table 1.4. Surgical margins play a significant role in local recurrence of disease (see Section 1.6.1). For distant disease or metastasis, however, the factor associated with the greatest relative risk (4.3) is histologic grade. The impact of local recurrence (LR) on survival remains an issue of debate. Randomised and non- randomised trials comparing amputation to limb sparing surgery or post operative radiotherapy, have failed to show any adverse effect of LR on survival (Pisters 2002). There is no doubt, however, that LR has a significant impact on the quality of life and surgical management of the patient.

Chapter 1: Soft tissue Sarcomas

1.7.2 Molecular markers

A large number of studies have carried out the search for newer molecular prognostic markers in STS. These have predominantly involved immunohistochemistry of variable numbers of archived paraffin-embedded tumour samples. Few studies have examined markers at transcript level (Table 1.5) and none of these employed a more global approach, such as investigating patterns of gene dysregulation. The first two studies listed in Table 1.5 are by the same group (Wurl, Kappler et al. 2002; Kappler, Kotzsch et al. 2003). They examined survivin, a member of the Inhibitor of Apoptosis (IAP) family (Altieri 2001) and TERT, the catalytic subunit of the enzyme telomerase present in >85% of epithelial tumours, which confers immortality to the tumour cells. The results confirmed an association between survivin expression and mortality. Co- expression of TERT and survivin was also shown. Telomerase is present in a proportion of soft tissue sarcomas but there is a greater prevalence of the alternate mechanism of telomere maintenance (ALT) in STS (up to 77% in malignant fibrous histiocytomas). ALT has not been shown as yet to correlate with histologic grade or survival in STS (Henson, Hannay et al. 2005).

Earlier work carried out at our laboratory (Oncology Research Centre, Prince of Wales Hospital, Sydney) investigated a number of potential prognostic markers. The expression of nm23 (NME2: non metastatic cells 2, protein expressed in) was evaluated in 46 STS specimens, revealing a positive correlation with tumour grade (D'Souza, Sheikh et al. 2003). Nm23 is a biomarker that had previously been proposed as a potential metastasis suppressor in various epithelial malignancies, including breast and ovarian cancer (Kauffman, Robinson et al. 2003). Nm23-H2 and cmyc were also overexpressed in STS (Surgical Oncology Research Group, unpublished data). The same samples were also evaluated for the expression of the insulin-like growth factor signal transduction pathway (Busund, Ow et al. 2004).

Chapter 1: Soft tissue Sarcomas

Table 1.4 Clinicopathologic Prognostic Factors for Extremity Soft tissue sarcoma 4#,2 ""4#01#0-%,-12'! !2-0 *-% ,) -6 /-! *0#!300#,!# 8%#1WR701 STXTRSX TRRSS /( 2.0#1#,2 2'-,TTRTRRRS TRRRS .-1'2'4#+ 0%',+'!0-1!-.'!STZTRRRS TRRRS "' 0-1 0!-+ TTWTRRRX TRRWZ .<3STZTRRRX TRRSR #2 12 1'1'8#WTRV[T[!+ST[TRRRS TRRRS '8#≥SRTR!+STWTRRRS TRTZ %'%&&0 "#VTUTRRRS TRRRS ##./-! 2'-,TTWTRRRS TRRRS /STYTRRRS TRTV <-,*'.-1 0!-+ &'12-*-%7RTXTRRRS TRRUS '1# 1#"0## '8#≥SRTR!+TTSTRRRS TRRRS 304'4 * ##./-! 2'-,TTZTRRRS TRRRT /( 2.0#1#,2 2'-,STWTRRUU /ST[TRRRS TRST .<3ST[TRRRS TRRYY .-1'2'4#+ 0%',+'!0-1!-.'!STYTRRSV TRSS /-5#0#620#+'271'2#STXTRRRS TRSX /(S /-! * 0#!300#,!#Q .<3S *'%, ,2 .#0'.� * ,#04# 1&# 2& 23+-30Q /S /#'-+7-1 0!-+ Q((S(#* 2'4#0'1)T/-%0 ,) ,"*-60#%0#11'-, 0##6.0#11#" 1.4 *3#1 .-**-!)TRRTT/-%V0 ,)5 131#" 12&##4#,21*'12#" 0# **2'+#"#.#,"#,24 0' *#1T 3&#*-5.4 *3#1',"'! 2#2& 22&#.-2#,2' *$-02&#! *!3* 2#"((12- #"3#2-!& ,!# 0# 4#07*-5 ,"2& 22#$ !2-01 0#',"#.#,"#,2*7.0-%,-12'!T3&#-0'%', *" 2 ..# 01', 123"7-$SRVS. 2'#,21 7.'12#01.F3#2 *QD**',6,!-*S[[XT

In particular, increased levels of IGFBP2 and IGF1-receptor beta (IGF1Rβ) correlated with grade on multiple regression analysis. Transforming growth factor β (TGFβ) and its receptors TGFβ-R1 and TGFβ-R2 were also found at increased levels in STS, compared to adjacent normal tissue and the expression levels correlated with the tumour grade (Unpublished data, Surgical Oncology Research Group, Prince of Wales Hospital; presented at RACS ASC, Brisbane 2003.)

The other studies of molecular markers have examined the following functional groups • cell cycle and apoptosis regulators, such as cyclins, cyclin dependant kinases (Kanoe, Nakayama et al. 1998; Goto, Kawauchi et al. 2003; Kawaguchi, Oda et al. 2003; Oliveira, Okuno et al. 2003), p53 (Pollock,

Chapter 1: Soft tissue Sarcomas

Lang et al. 1996; Konomoto, Fukuda et al. 1998; Medina-Franco, Ramos-De la Medina et al. 2003), mdm2 (Kanoe, Nakayama et al. 1998), bad, bax, bcl-2 (Kohler, Wurl et al. 2002) and nm23 (D'Souza, Sheikh et al. 2003), • growth factors and receptors such as EGFR and IGF-1 (Beech, Pollock et al. 1998; Merimsky, Issakov et al. 2002; Nuciforo, Pellegrini et al. 2003; Busund, Ow et al. 2004; Thomas, Giordano et al. 2005) (see section below), • proteases and factors involved in invasion and metastasis (Benassi, Gamberi et al. 2001; Aryee, Ambros et al. 2002) and • angiogenic factors (Chao, Al-Saleem et al. 2001; Yudoh, Kanamori et al. 2001; Potti, Ganti et al. 2004)

What is clear from the overall review of the studies to date, is that functional groups of factors, such as those noted above, play a role in tumorigenesis and progression of disease. Rather than focusing solely on individual markers, in reductionist fashion (Sporn 1996), it may in fact further our understanding of tumour biology if a more global approach were employed. The molecular biology of soft tissue sarcoma is therefore discussed in this context, in the following sections.

Chapter 1: Soft tissue Sarcomas

Table 1.5 Studies on Molecular Prognostic markers in STS, 1996-2005 "32&-0 F-30, * 0-%,-12'!# 0)#0 F30*. / ,!#2′RT 304'4',#& 4# ,',4#01#!-00#* 2'-,5'2&1304'4 * 7 ..*#0 **',* (#1′RU 304'4',& 1 ,',4#01#!-00#* 2'-,5'2&1304'4 * #, 11' 8,, *16,!-*-%7′RS ↑.T.[Q↓3.T!-00#* 2#5'2&+#2 12 1'1 ##!& ,2D6,!′[Z -&"(&"VS',U3 313,"/3 :'0!&80!&′RV ↑&".T&"S(β!-00#* 2#5'2&23+-30%0 "#',VX3 *& -* 8,, *130%6,!′RS :-&"!-00#* 2#15'2&23+-30%0 "# 807## <3 . #"(#1′RT ↓78VS * ZT',-5',%1< 0#',&-$#07- ,2D* ′[V .*<8!-00#* 2#15'2&%0 "#',"% 0#',&-$#07- *72-+#207′RT <8.*-'"7,-2 .0-%,-12'!$ !2-0',[U3 ′-38 (D 8,2'! ,!#0(#1′RU ↑,+TU!-00#* 2#15'2&23+-30%0 "#',VX3 8'#%'#*. %'12-%'12-. 2&′RW #2 **-2&'-,#',7'VXY!-00#* 2#5'2&%0 "#1304'4 * &-2-? * !'#,!#′RU ↓.TY↑*7!*',-!-00#* 2#5'2&+#2 12 1'1↓1304'4 * &3#00#0- "8-′RT )0 1+32 2'-,1',3+-31#+-"#*1 7 ,-#% 8,2'! ,!#0(#1′[Z ↑$ZQ XQ ',3 7 5 %3!&'7 D. 2&′RU ↓.SX!-00#* 2#15'2&.--0.0-%,-1'1',YY/ 7-&*#03 8,2'! ,!#0(#1′RT ↑ "Q 6Q !*VXQ !*V6+!-00#* 2#5'2&.--0.0-%,-1'1 7-,-+-2-3 %3+ ,. 2&′[Z .[Y+32 2'-,1!-00#* 2#5'2&.--0.0-%,-1'1',/ #"', V"0 ,!-% 8,, *130%6,!′RU ↑.WU',VY&1 0!-+ 1!-00#* 2#5'2&.--0.0-%,-1'1 .-**-!)(- 6,!-%#,#′[X **-, *#6. ,1'-,-$+32 2#".[Y 11-!' 2#"5'2&+#2 12 1#1 /'H; . 2&,2*′RU -(αβ#6.0#11#"',3T<-.0-%,-12'!',$-0+ 2'-, #0'+1)76 8!2 6,!-*′RT <-#0 T-4#0#6.0#11'-,',TUR3 #0'+1)76 -30D* ′RT ↓#0 V.-12!&#+-2� .7',0#1.-,1'4#. 2'#,21 <3!'$-0-.& %3+ ,. 2&′RU ↑#0VX ,#3 11-!' 2#"5'2&%--".0-%,-1'1',SU 3&-+ 1 & * ,!#0′RW -&"(%-(VT ,#3#6.0#11#" 2*-5*#4#*1', 6*'4#'0 8 D*6′RU ).T!-00#* 2#15'2&.--0.0-%,-1'1',SZT3 '2-3 6,!-%#,#′RV V! "�',+32 2'-,↑, '*',VR 3 3 #02%F * ,!#0′RU ↓& X!-00#* 2#15'2&.--0.0-%,-1'1',WY3 .-22'8 D* (#1**',6,!′RV :-&"!-00#* 2#15'2&1304'4 *',/ ?3"-&7 D*′RS:-&"!-00#* 2#15'2&%0 "#.--0.0-%,-1'1',SSW3 3S -$2 2'113# 1 0!-+ Q S 7,-4' * 1 0!-+ Q "%S *'%, ,2 $' 0-31 &'12'-!72-+ Q /S /#'-+7-1 0!-+ Q <S <#30- * 12-+ T -&"(S -.'"#0+ * %0-52& $ !2-0 0#!#.2-0Q -(S 6#120-%#, 0#!#.2-0Q &"S ,13*',V*')# %0-52& $ !2-0Q &"S(βS ,13*',V*')# %0-52& $ !2-0 S 0#!#.2-0 #2 Q &".TS ,13*',V*')# %0-52& $ !2-0 ',"',% .0-2#', TQ ,+TUS ,-, +#2 12 2'! !#**1 TQ .0-2#', #6.0#11#"',Q.*<8S.0-*'$#0 2',%!#**,3!*# 0 ,2'%#,QQ3-(3S3#*-+#0 1#0#4#01#20 ,1!0'.2 1#Q .S 20'6+#2 **-.0-2#', 1#Q3.S3'113#',&' '2-0-$+ 20'6+#2 **-.0-2#', 1#T

Chapter 1: Soft tissue Sarcomas

1.7.3 EGFR as a Prognostic Marker

EGFR (HER1/ErbB1) is a ligand-activated transmembrane receptor tyrosine kinase that is expressed in many non-neoplastic tissues. It is known to be involved in the development and differentiation of neurons and astrocytes, keratinocytes and other epithelial tissue, chondrocytes and osteoblasts (Sibilia, Kroismayr et al. 2007). Once ligand binding by molecules such as epidermal growth factor, transforming growth factor alpha, amphiregulin, epiregulin, among others, has occurred, homo- or heterodimerisation causes phosphorylation of tyrosine residues, leading to activation of downstream signaling cascades (Yarden 2001; Barnes and Kumar 2003; Scaltriti and Baselga 2006)2. EGFR is thought to play an important role in cellular proliferation, survival and migration (Zandi, Larsen et al. 2007).

Dysregulation of EGFR has been implicated in many epithelial malignancies, with up to 100% of squamous cell cancers of the head and neck overexpressing EGFR. EGFR involvement has also been noted in other epithelial malignancies such as pancreatic, renal cell, colorectal, breast, ovarian, prostate, bladder and lung cancer (Freier, Joos et al. 2003; Hirsch, Varella-Garcia et al. 2003; Mendelsohn and Baselga 2003; Thomas, Chouinard et al. 2003; Ciardiello, De Vita et al. 2004; Dawson, Guo et al. 2004; Baselga and Arteaga 2005; Hynes, Lane et al. 2005; El-Rayes, Ali et al. 2006; Sharma, Bell et al. 2007).

The mechanisms by which tumorigenesis occurs have been postulated to include increased production of ligands, EGFR gene amplification, mutations in the EGFR gene resulting in constitutive activation, overexpression of the EGFR protein, ErbB2/EGFR heterodimerisation, reduced receptor downregulation and cross talk between receptor systems (Zandi, Larsen et al. 2007). Overexpression of EGFR without gene

2 Activated EGFR and its downstream signalling cascades, in particular the Ras-MAPK, PI3K-Akt and Jak-STAT pathways, are discussed and examined further in Chapter 4 of this thesis. 46

Chapter 1: Soft tissue Sarcomas

amplification may occur as a result of increased activity of the EGFR promoter or abnormalities at the translational and post-translational levels. Mutations causing constitutive activation of EGFR or overexpression of EGFR can lead to spontaneous dimerisation, independent of ligand binding, and consequent activation of downstream pathways. EGFR may also be associated with tumor progression by triggering angiogenesis and bone destruction (De Luca, Carotenuto et al. 2008).

The mechanism of activation of EGFR and its pathways has implications for the efficacy of targeted therapy. Better response rates to small molecule tyrosine kinase inhibitors or monoclonal antibodies have been noted in patients exhibiting increased EGFR copy number and/or mRNA expression (Hirsch, Varella-Garcia et al. 2003; Moroni, Veronese et al. 2005). This is discussed further in Chapters 4 and 8.

The evidence for EGFR involvement in sarcoma oncogenesis and tumour progression has been less prolific. Some of these studies have had conflicting results, such as those investigating the prognostic role this marker might play. EGFR expression in synovial sarcomas has been reported in gene expression studies (Allander, Illei et al. 2002; Nielsen, West et al. 2002; Baird, Davis et al. 2005). Another gene expression study did not, however, report an overexpression of EGFR in synovial or other sarcomas (Nagayama, Katagiri et al. 2002). One study has examined protein overexpression of EGFR in a variety of STS (Sato, Wada et al. 2005) and is discussed further in Chapter 4 (Section 4.4.2) in relation to the present thesis. Other mechanisms by which EGFR may be oncogenic, or contribute to tumour progression within STS has not been fully evaluated, although EGFR mutations were not found to be present in one recent study (Baird, Davis et al. 2005). The mechanism of EGFR signal transduction within STS, has not, to our knowledge, been investigated.

EGFR as a potential prognostic marker of tumour progression within STS and its signal transduction is examined in this thesis.

1.8 BIOLOGY of SOFT TISSUE SARCOMAS

Three concepts are central to forming an overarching understanding of malignant transformation of mesenchymal tissue. The first deals with mesenchymal cell

Chapter 1: Soft tissue Sarcomas

differentiation, the second, with chromosomal anomalies in STS and the third, with telomere maintenance mechanisms (TMM) in sarcoma.

1.8.1 Cell Lineage Differentiation

In general, soft tissue sarcomas do not seem to result from malignant changes or the dedifferentiation of benign soft tissue tumours. A pluripotent cell is thought to give rise to the main mesenchymal cell lineages, namely, fibroblasts, chondroblasts, osteoblasts, lipoblasts and muscle as depicted below.

Figure 1.2 Key genes in lineage differentiation from the pluripotent mesenchymal stem cell MYOD and myogenin (MYOG) are required for skeletal muscle differentiation, peroxisome proliferator-activated receptor gamma 2 (PPARγγγ2) is responsible for adipocyte differentiation, CBFA1 determines osteoblastic differentiation and SOXG is integral to cartilage formation. Key regulators of fibroblastic differentiation are yet to be elucidated. This figure was reproduced from Pollock, 20023.

3 The original figure appears in a review by Rodan and Harada where the “master genes” responsible for lineage specific differentiation are discussed (Rodan and Harada 1997) 48

Chapter 1: Soft tissue Sarcomas

The cell lineages form the basis of much of the histologic classification of soft tissue tumours where the various subtypes are named according to the presumed cell of origin. Difficulty arises, however, when the cell of origin cannot be determined, such as in Ewing’s sarcoma, clear cell sarcoma or alveolar rhabdomyosarcoma (ARM). These are examples of where “phenotypic switch” (mesenchymal to epithelial) may have occurred, where a Ewing’s tumour may express cytokeratin, which is usually associated with epithelial differentiation. Just as epithelial-mesenchymal transition (EMT) has been recognised as playing a role in the dedifferentiation and progression of epithelial tumours4, via signal transduction pathways that involve tyrosine kinases and transforming growth factor beta (TGF!), the converse is also possible. Mesenchymal to epithelial transition occurs during organogenesis of the kidney (Thiery 2002), triggered by genes such as Wilm’s tumour 1 (WT1) and WNT4. Phenotypic switch, therefore, has to be considered, not only in the classification of soft tissue sarcomas, but in the fundamental conceptualisation of STS pathogenesis.

1.8.2 Cytogenetics: Simple and Complex

Integral to our overall understanding of the molecular pathogenesis of soft tissue sarcomas has been the recognition that cytogenetic aberrations play a pivotal role. In terms of cytogenetics, sarcomas fall into two broad categories: those with simple karyotypes, often with a characteristic single recurring translocation, and those with complex karyotypes, a hallmark of their inherent genomic instability (Ladanyi and Bridge 2000; Skapek and Chui 2000; Tomescu and Barr 2001; Helman and Meltzer 2003; Hogendoorn, Collin et al. 2004; Mandahl, Mertens et al. 2004).

4 Epithelial-mesenchymal transition (EMT) has been known to be involved in embryogenesis since 1982. The role it plays in tumour progression has been more recently defined and is reviewed by Jean Paul Thiery (Thiery 2002). Others have discussed the involvement of factors such as Snail (Cano, Perez- Moreno et al. 2000), TGF! (Yu, Hebert et al. 2002; Zavadil, Cermak et al. 2004) and Notch (Cano, Perez-Moreno et al. 2000; Yu, Hebert et al. 2002; Grego-Bessa, Diez et al. 2004; Timmerman, Grego- Bessa et al. 2004) in this process. 49

Chapter 1: Soft tissue Sarcomas

1.8.2.1 Sarcomas with simple karyotypes

Mesenchymal tumours with relatively simple cytogenetic aberrations include those that are more commonly seen in childhood such as the Ewing family of tumours and alveolar rhabdomyosarcomas (ARMS), as well as others such as synovial sarcomas and myxoid liposarcomas. They are characterised by single recurrent chromosomal translocations that tend to be present in a significant proportion of the tumours, as summarised in Table 1.6.

In some cases, such as with synovial sarcomas and Ewing’s sarcoma, there may be more than one potential fusion transcript that can result from the reciprocal translocation. There is typically a common breakpoint with a number of alternative translocations in these situations, but the resultant tumours are generally indistinguishable on histology. An exception to this is the EWSR1 breakpoint which is involved in Ewing’s sarcoma, as well as translocations that result in myxoid liposarcoma and clear cell sarcoma, among others and the FUS gene, which also forms fusion transcripts with a number of genes to result in morphologically different tumours (detailed in Figure 1.3). It is interesting to note that the different transcripts may have prognostic significance. For instance, the SYT-SSX1 fusion in synovial sarcoma is associated with biphasic histology and worse prognosis than SYT-SSX2. In addition, different transcripts may exist within one tumour as a result of alternative splicing.

Most of these fusion genes encode transcription factors and result in dysregulation of downstream genes and pathways. The detailed downstream mechanisms affected by these fusion oncoproteins has not been elucidated in all cases, particularly where the “normal” cell counterpart of the tumour subtype is unclear, such as in Ewing’s sarcoma.

Given that these translocations are specific for their respective tumour types, the nature of this association has naturally provoked great interest. It may be that the chromosomal rearrangements are cell lineage-specific (Barr 1998). Alternatively, these recurrent translocations occur in an undifferentiated precursor cell, which would in turn imply that the gene fusion, in effect, selects the lineage. Either concept is tenable. Moreover, it follows that the translocations occur early in the genesis of the tumours, determining the tumour phenotype.

Chapter 1: Soft tissue Sarcomas

The process by which these translocations occur also remains to be clarified. It is thought that the translocation is either a random event, such as nonhomologous end joining of DNA (NHEJ)5, or a specific DNA recombination event (Skapek and Chui 2000). Analysis of the genomic DNA sequences near the breakpoints in the t(11;22) translocation of Ewing sarcoma (Obata, Hiraga et al. 1999) and the t(12;16) translocation of myxoid liposarcoma (Kanoe, Nakayama et al. 1999) found DNA- binding sites in these flanking regions, indicating a role for these DNA-binding domains in the translocation process.

5 The NHEJ pathway and its relevance for genomic stability is discussed in greater detail by Sharpless and colleagues (Sharpless, Ferguson et al. 2001) 51

Chapter 1: Soft tissue Sarcomas

Table 1.6 Fusion Genes in Sarcomas 0!-+ 7.# 0 ,1*-! 2'-, 31'-,#,# 0#4 *#,!#$ - .<-3 2SSRTT/TVR/ST .V/+W [W - .<-3 2TSRTT/TTR/ST .V# W - .<-3 2YRTT.TTR/ST .V0W 0S - .<-3 2SYRTT/STR/ST .VW / 0S - .<-3 2TRTT/UUR/ST .V/0 0S (*3 2SSRTT.SUR/ST .V.W 1[[ 76-'"/. 2STRSX/*UR.SS +V$1 [W 76-'"/. 2STRTT/SUR/ST .V$1 0W -* 2[RTT/TTR/ST .V$2 YW -* 2[RSY/TTR/SS /X2V$2 TW . ** 2STRTT/SUR/ST .V /W # 7,-4' *1 0!-+ 2HRSZ.SSR/SS 3V4W XW 7,-4' *1 0!-+ 2HRSZ.SSR/SS 3V4X UW 7,-4' *1 0!-+ 2HRSZ.SSR/SS 3V4Z # 8*4#-* 0( 2TRSU/UWR/SV 4YV/# YW 8*4#-* 0( 2*RSU.UXR/SV 4]V/# SR ". 2SYRTT/TTR/SU $1+W WV /## 1[[ *-,%#,'2 *" 2STRSW.SUR/TW 0\V2#Y 1[[ - 2YRSY.SWR/TS 7 8/WV77 8W # -Q-5',%11 0!-+ R.<-3Q.#0'.�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

Chapter 1: Soft tissue Sarcomas

EMC CCS AFH

TCF12, 15q21 NR4A3 TAF15, 17q11 9q22 ATF1 12q13 FUS BBF2H7 7q33 FLI1, 11q24 DDIT3 16p11

HSRNAFEV, 2q33 12q13 ETV1, 7p22 EWSR1 E1AF, 17q12 MLS LGFS ZNF378, 22q12 20q12 ERG WT1 21q22 11p13

ES DSRCT

SS18L1 SSX1 PAX7 20q13 Xp11 FOXO1A 1p36 (FKHR) SSX2 PAX3 13q14 Xp11 2q35 SS18 SSX4 18q11 Xp11

SS ARMS

Figure 1.3 Translocations in cytogenetically simple soft tissue sarcomas The ellipses indicate the resultant tumour types. The boxes contain the genes and their chromosomal locations. The double arrows indicate the translocations. Some genes are involved in multiple translocations resulting in morphologically different tumours, such as EWSR1. AFH: Angiomatoid fibrous histiocytoma, ARMS: Alveolar Rhabdomyosarcoma, CCS: Clear cell sarcoma, DSRCT: Desmoplastic small round cell tumour, EMC: Extraskeletal myxoid chondrosarcoma, ES: Ewing’s sarcoma, LGFS: Low grade fibromyxoid sarcoma, MLS: Myxoid liposarcoma, SS: Synovial sarcoma. This figure was adapted from Mandahl, 2004 (Mandahl, Mertens et al. 2004).

Chapter 1: Soft tissue Sarcomas

Studies have also attempted to examine the downstream effects of the fusion oncoproteins. Transgenic mouse modeling of these fusion transcripts have had limited success, apart from one utilising the chimeric FUS/TLS-CHOP associated with myxoid liposarcomas (Perez-Losada, Pintado et al. 2000). This study was able to demonstrate that the transgenic mice exclusively developed liposarcomas and furthermore, that the tumours expressed high levels of peroxisome proliferator-activated receptor gamma (PPARγ), which is required for adipocyte differentiation. This allowed the authors to propose that the “differentiation block” of the fusion oncoprotein on the pluripotent mesenchymal cell occurred downstream of PPARγ but they did not go on, in this paper, to elucidate the involved mechanisms.

DNA microarrays have the potential to provide hypotheses for the potential downstream targets of fusion transcripts. In an in vitro study (Khan, Bittner et al. 1999) of paediatric alveolar rhabdomyosarcoma (ARMS), where Pax3 and the chimeric transcript PAX3- FKHR were introduced into NIH 3T3 cells, the resultant gene expression profiles were compared to those of established ARMS cell lines. 11 genes were found to be induced in the PAX3-FKHR-transduced lines and 4 genes that were repressed. These were all muscle-related genes. Using less stringent criteria, the number of induced genes was 111, 28 of which are involved in muscle transcription. Signal transducers such as IGF2 and IGFBP5 which are known to be involved in muscle growth and differentiation during embryogenesis, were also induced.

The cause and downstream effects of the reciprocal recurrent translocations are, with advances in molecular techniques, being elucidated. While the prognostic significance of the translocations may not always be known, their diagnostic relevance is appreciated and has already been put into practice. Cytogenic characterisation can be carried out on fine needle aspirates, with fluorescence in situ hybridisation used to detect the fusion genes. Reverse transcriptase polymerisation can also be employed to detect the translocations using paraffin-embedded archival material (Guillou, Coindre et al. 2001).

Chapter 1: Soft tissue Sarcomas

1.8.2.2 Sarcomas with complex karyotypes

The more common sarcomas which generally affect an older age group fall into the category of tumours with complex chromosomal aberrations. These include the pleomorphic liposarcomas, fibrosarcomas, malignant fibrous histiocytomas, malignant peripheral nerve sheath tumours and leiomyosarcomas. Numerous unbalanced translocations can occur in these tumours, with marked differences between cells in the same tumour. The underlying genomic instability can lead to losses of various chromosome regions. Gene amplifications also tend to occur in this group of tumours, manifesting cytogenetically as double minute chromosomes and homogeneous staining regions. Some of these losses and gains are summarised in Table 1.7, where the most frequent aberrations in the tumours studied are mentioned.

As shown in Table 1.7, a number of studies focused on malignant fibrous histiocytomas (MFH) (Mairal, Terrier et al. 1999; Parente, Grosgeorge et al. 1999; Sakabe, Shinomiya et al. 1999), using either conventional cytogenetics or comparative genomic hybridisation (CGH). Apart from loss of chromosome region 2q32, there was little else in common between the studies.

Loss in chromosome region 13q was a common thread among studies investigating leiomyosarcomas (LMS) (Parente, Grosgeorge et al. 1999; Mandahl, Fletcher et al. 2000; Otano-Joos, Mechtersheimer et al. 2000; Derre, Lagace et al. 2001). Loss of regions 1p and 9p were noted in two separate studies examining MPNST (Fletcher, Dal Cin et al. 1999; Mechtersheimer, Otano-Joos et al. 1999) but not in a third study of 31 MPNSTs (Schmidt, Wurl et al. 1999).

Gastrointestinal stromal tumours (GIST) tend to be more uniform in their cytogenetic characteristics. The findings are usually less complex, involving deletion of chromosome 14, 22 or 1p. High grade GISTS consist of at least 3 of these abnormalities (Andersson, Sjogren et al. 2002; Gunawan, Bergmann et al. 2002; Heinrich, Rubin et al. 2002). A study of 95 GISTs reported finding, on average, 2.6 chromosomal aberrations in benign GISTs, 7.5 in malignant primary GISTs and 9 in metastatic GISTs (El-Rifai, Sarlomo-Rikala et al. 2000).

Chapter 1: Soft tissue Sarcomas

Sarcomas with complex karyotypes are often associated with mutations in the tumour- suppressor genes TP53 and RB1, indicating that loss of DNA-repair mechanisms may underpin the genomic instability of these tumours. Despite the greater frequency of p53 mutations in these types of sarcomas, the prognostic significance appears to be less than that for p53 mutations that occur in sarcomas with simple karyotype, such as Ewing’s sarcoma (Wurl, Meye et al. 1999; Antonescu, Leung et al. 2000; de Alava, Antonescu et al. 2000).

Another major cause of complex chromosomal aberrations in these tumours may be telomere dysfunction. Telomeric dysfunction can lead to unstable ring and dicentric chromosomes and anaphase bridges. It can cause chromosomal fragmentation mediated by persistent bridge-fusion-breakage (BFB) events leading to a continuous reorganisation of the tumour genome (Gisselsson, Jonson et al. 2001). Whole chromosome losses could also occur in this manner. The maintenance of telomeres by telomerase or alternate mechanism (ALT)6 may then play a role in preventing cell death, preserving the tumour cell with its disordered genome.

A number of mechanisms are clearly involved and the current interest in cytogenetic analyses of soft tissue sarcomas with complex chromosomal aberrations will aid in unraveling the biologic processes. At present, it is not known which of the many chromosomal changes in these tumours are central to the tumour biology. Nonetheless, some understanding of the prognostic significance of the cytogenetic changes has been gained.

6 The alternate lengthening of telomeres mechanism (ALT) is discussed briefly in the section on genetic syndromes (Li Fraumeni syndrome) associated with predisposition to STS and in greater detail in the following section on telomere maintenance mechanism (TMM), as well as in Chapter 5 on characterisation of the LMS-LFS cell line. 56

Chapter 1: Soft tissue Sarcomas

Table 1.7 Cytogenetic aberrations in soft tissue sarcomas of complex karyotype 832&-0 ?# 0 33+-3037.# *72-%#,#2'! ,-+ *'#1 #02#,1" S[[Z VZ. &10 "+',R0# 00 ,%#+#,21-$S[.SU S[/SU',&'%&%0 "#23+-301 '0 *8 S[[[ UR"% /-11#1SSU/STV/SV/TSQSS/TUQT/UTQ SS.SUQSR.QS/VQ[.TQSX/STQV/UQSR/TWQ U.TUQT.TVST. . 0#,2#" S[[[ X[3 /S/-11#1SS/SVV/2#0SU/TSV/TT "%S/-11#1T.TVV.2#0T/UTV/2#0 "*#2!�* S[[[ VS/QTY.<3 /S/-11#1S.QSS/QST/QSR. .<3S/-11#1S.Q[.QSR.QSS/QSY.QSY/ ) #3 S[[[ S["% & ',1SV/STVTSQZ.TSV.2#0QZ/TVTSV/2#0Q [/STVSUQST.SSTTV.2#0SW/SSTTVSW #!&2#01&#'+#0& S[[[ S[.<3 /-11#1S[.TSV.TVQSU/SVV/TTS. & ',1SSY/TVV/TWQY.SSV.SUQW.SWQ Z/TTV/TVST/TSV/TV !&+'"2% S[[[ US.<3 /-11SSV/TVTUV/2#0 & ',1SZ/TUV/TVTSQZ/TSTSV/TTQ Z/TVTTV/2#0W.SVQX.TTV.2#0QY.SWV.TSQ Y/UTV/UWSY/TTV/2#0 62 ,-VD--1 TRRR SV/ & ',1SW.SWQZ/TVQSW/TWVV1/TXQSY.QH. /-11#1SSR/SU/ ," &*< TRRR VW/ /-11#1SU.TSV.TUQZ.TSV.2#0QSU/STV/SUQ SU/UTV/2#0QS/VTV/2#0QT.SWV.2#0QSZ.SSQ S.UXQSS/TUV/2#0SR/TUV/2#0 & ',SS/STV/US #00#D TRRS TY/ & ',1SS.UQS/TSQSW/STVSWQSX.QSY.Q SY/QS[QTR/QTT/QH.T /-11#1SSU/SVVTSQS/QT.T/QV/Q[.QSR. SR/QSS.SS/TUSX/ !&+'"2% TRRT VS" & ',1SST/QSV/TTV/TVQV/TSV/TVQ Y/USZ/TT F ,%( TRRU TZ/ /-11SSU/SVV/TSR& ',SW.SVV.2#0 8,"#011-,D TRRT SR&3 /-11#1S*&0-+-1-+#1SVTT &3, 5 , TRRT S[&3 /-11#1SSVQTTQ[.QS.QSWQU.QSU/QSR/QS[ #02#,1" TRRT SWS3Q&'%&%0 "# 0# ).-',2SS.SQS/VQSV/SQSY/TT & ',1SX.S .T "S"' 0-1 0!-+ 1Q&3S& 120-',2#12', *120-+ *23+-30Q/S/#'-+7-1 0!-+ 1Q"%S *'%, ,2 $' 0-31 &'12'-!72-+ 1Q .<3S *'%, ,2 .#0'.� * ,#04# 1&# 2& 23+-301Q .S .*#-+-0.&'! 1 0!-+ 1Q 3S -$2 2'113# 1 0!-+ 1Q "+',S "-3 *# +',32# !&0-+-1-+#1 ," &10S &-+-%#,-31*7 12 ',',%0#%'-,1 0# 11-!' 2#"5'2& <8 +.*'$'! 2'-,1T3&# -4#"-#1,-20#.0#1#,2 , #6& 312'4#*'12-$!72-%#,#2'!123"'#1',3-$!-+.*#6) 07-27.#T

Chapter 1: Soft tissue Sarcomas

In a study of 151 patients with high-grade sarcoma, five independent predictors of metastasis were identified, namely, breakpoints in chromosome regions 1p1, 1q4, 14q1, and 17q2, and gain of regions 6p1/p2 (Mertens, Stromberg et al. 2002). A study of 28 LMS found survival time to decrease with an increase in the number of chromosomal aberrations identified (Wang, Titley et al. 2003). In terms of specific aberrations associated with prognosis, 13q14-q21 loss and 5p14-pter gain at diagnosis identified patients likely to have a shorter survival time. The only report to date investigating the prognostic relevance of cytogenetic anomalies in fibrosarcomas (Schmidt, Taubert et al. 2002) found gain of 12q22 to correlate significantly with a poor overall survival rate.

Several factors seem clear, despite the complexity of these karyotypes: • certain aberrations are common to tumours of the same histologic subtype, • tumours acquire chromosomal anomalies as they progress, an indication of their inherent genomic instability and aggressiveness and • prognostic information can be obtained from chromosome analysis. As more studies are carried out with larger sample sizes, more chromosomal prognostic markers may be identified and ultimately come to be used as therapeutic targets. The correlation of chromosomal losses and gains with specific gene amplifications and deletions may add important new cellular targets for future therapies.

1.8.3 Telomere Maintenance Mechanisms (TMM)

The telomere phenotype of soft tissue sarcomas is a subject of current interest. Telomeres are the protective caps on the ends of chromosomes characterised by TTAGGG repeats. In mortal cells, there is progressive loss of the telomeres with each cell division by 40-200bp, leading to senescence at 5-8 kb. It is thought that senescence may be a barrier against tumourigenesis (Reddel 2000). Immortalised cells, tumours and some normal human embryonic cells maintain their telomere lengths and do not senesce. These immortal cells possess a telomere maintenance mechanism (TMM). Previous surveys of mostly epithelial cancers have shown that 85% tumours utilise telomerase as their TMM (Shay and Bacchetti 1997) and maintain their telomere length. The catalytic subunit of telomerase is a reverse transcriptase, TERT (Kilian, Bowtell et al. 1997).

Chapter 1: Soft tissue Sarcomas

It has been shown, however, that compared to epithelial cancers, a greater proportion of sarcomas, do not possess telomerase and utilise an alternate mechanism for lengthening of telomeres (ALT). ALT is a phenomenon characterised by ALT-associated PML (promyelocytic leukaemia) bodies or APBs and heterogeneous telomere lengths (Mean 20 kb; Range 3-50 kb) (Bryan, Englezou et al. 1995; Yeager, Neumann et al. 1999; Reddel 2003; Tsutsui, Kumakura et al. 2003). A recent review showed a large proportion of sarcoma cell lines and cell lines derived from breast stroma and fibroblasts of patients with Li Fraumeni syndrome are ALT+ (Henson, Neumann et al. 2002). One study noted that telomerase was present in sarcomas of simple karyotype but those with complex karyotypes utilised ALT (Montgomery, Argani et al. 2004). A larger study found prevalence of ALT in sarcomas to range from 6% to 77%, depending on the histologic subtype (Henson, Hannay et al. 2005). 50% of high grade STS and 33% of metastases were ALT+.

ALT is more often present in tumours of mesenchymal, rather than epithelial origin. It is possible that there is, as speculated by Henson and colleagues, greater repression of telomerase in mesenchymal cells with a relative greater probability of activating ALT (Henson, Neumann et al. 2002). Mesenchymal cells have a slower turnover than epithelial cells and potentially undergo less telomere shortening for this reason. Loss of p53 function being an early event in Li Fraumeni syndrome may contribute to tumour cells from these patients being ALT+. Some tumours may have both ALT and telomerase, as demonstrated in a study examining liposarcomas (LPS) (Johnson, Varkonyi et al. 2005). This may be due to the presence of different sub population of cells, some with telomerase, some with ALT or possibly the activation of both mechanisms in every cell.

Telomerase and ALT may contribute to carcinogenesis in following ways: • Maintenance of telomere length and escape from senescence. • Excessive telomerase or ALT activity in otherwise normal tissue at some stage during development with increased reserves of telomere length allowing clonal expansion of these cells once oncogenic mutations have occurred in them

Chapter 1: Soft tissue Sarcomas

• Involvement in abnormal repair events such as end-end fusions which cause chromosomal breaks and genomic instability. The theory that ALT and genomic instability are linked was recently investigated in a study on liposarcomas (LPS) (Johnson, Gettings et al. 2007). ALT+ LPS demonstrated more copy number changes than telomerase positive LPS, or those without either TMM. Most of these copy number changes were amplifications, although deletion of chromosome 1q32.2-q44 was also found to be frequent. No prognostic correlation was carried out however, in this study or in the authors’ previous study (Johnson, Varkonyi et al. 2005). Another study, also concentrating on LPS, did however find a correlation between the presence of TMM and an unfavourable outcome, with the more aggressive dedifferentiated LPS being more frequently ALT+ (Costa, Daidone et al. 2006).

In general, the prognostic significance of the TMM in sarcoma is yet to be firmly established. Further characterisation of TMM in soft tissue sarcoma will play a role in the development of new therapeutic strategies.

1.9 MOLECULAR BIOLOGY and RELEVANCE for TARGETED THERAPY

The preceding section outlined the genomic aberrations that can occur in soft tissue sarcomas. The sarcomas of simple karyotype, which tend to be paediatric sarcomas, are characterised by recurrent translocations that typically result in fusion oncoproteins. On the other hand, multiple clonal anomalies with gene amplifications and deletions are the hallmark of the sarcomas of complex karyotype. Genomic instability, possibly due to the loss of protective tumour suppressor genes is the underlying trait in these cases. The prognostic value of molecular cytogenetics is being realised. The ultimate goal, however, is to identify potential therapeutic targets.

1.9.1 Fusion Oncoproteins

The majority of fusion oncoproteins encode transcription factors (Table 1.6). Inactivation of these factors would inhibit the early growth and progression of tumours. In liposarcomas (LPS), for instance, potential targets include peroxisome proliferator- activated receptor (PPARγ), required for adipocyte differentiation. Giant marker or ring chromosomes, present in well differentiated LPS are potential targets, as is the TLS-

Chapter 1: Soft tissue Sarcomas

CHOP fusion product in myxoid LPS. The PPARγ- ligand troglitazone (Demetri, Fletcher et al. 1999) and rosiglitazone (Debrock, Vanhentenrijk et al. 2003) has been used in clinical trials for the treatment of LPS, with the aim of inducing lineage- appropriate terminal differentiation in the malignant cells. The ring chromosomes typically contain material from chromosome 12, with gene amplifications. These amplifications, once fully identified could form the basis for targeted therapy.

Fusion oncoproteins are potentially immunogenic: Their peptide products would need to be presented on the cell surface in conjunction with major histocompatibility complex (MHC) Class I molecules or antigen presenting cells for immune recognition. Murine models and in vitro studies suggest that tumor cells can be lysed by cytotoxic T cells of the immune system provided the T-cells have been primed by antigen-presenting cells. (Tomescu and Barr 2001). Generation of vaccines directed against the oncoproteins therefore seems logical. Early trials of vaccines against the Ewing sarcoma gene EWS, have not however been successful (Fletcher 2004) and further development of these vaccines are required. In a small Phase 1 trial of a junction peptide directed against the SYT-SSX fusion oncoprotein in synovial sarcoma, 6 patients with disseminated disease received the vaccinations (Kawaguchi, Wada et al. 2005).

1.9.2 Tyrosine Kinases

A number of receptor tyrosine kinases (RTKs) have been implicated in the pathogenesis of STS. These include activating mutations of KIT and platelet derived growth factor A (PDGFRA) in Gastrointestinal stromal tumours (GIST) (Duensing, Heinrich et al. 2004), the fusion oncoprotein COL1A-PDGFB with overexpression of PDGFB in Dermatofibrosarcoma protuberans (DFSP) (Fletcher 2004), insulin-like growth factor 1 receptor (IGF1R) in embryonal rhabdomyosarcomas and epidermal growth factor receptor (EGFR) in synovial sarcomas (Allander, Illei et al. 2002; Blay, Ray-Coquard et al. 2004; Albritton and Randall 2005). EGFR expression and its involvement in tumour progression in STS are explored in this thesis.

The success of targeted therapy is exemplified by the use of imatinib mesylate (STI 571/Gleevec®/Glivec®, Novartis, Basel, Switzerland) in the treatment of GISTs (van Oosterom, Judson et al. 2001; van Oosterom, Judson et al. 2002). Imatinib, which

Chapter 1: Soft tissue Sarcomas

inhibits the KIT receptor by binding at its tyrosine kinase domain7, has achieved partial response or stable disease in as many as 89% of patients with unresectable metastatic GISTs (Scappaticci and Marina 2001). This has spurred further trials examining the efficacy of imatinib in the treatment of other STS (Verweij, van Oosterom et al. 2003. The importance of RTKs in the oncogenic mechanisms has been highlighted and researchers have been steered to investigate the role of other RTKs in STS. To this end, other agents such as the EGFR inhibitor ZD1839, have been trialled for the treatment of synovial sarcoma {Blay, 2004 #494). EGFR-directed therapy is discussed in greater detail in Chapters 4 and 8.

1.9.3 Tumour suppressors

Loss of p53 and RB1, both of which function as tumour suppressors, has been discussed in the preceding sections as being frequently noted in STS. Cells with DNA damage continue to replicate as apoptosis is not triggered.

Adenovirus E1A proteins bind and neutralise RB, releasing the transcription factor E2F, which is repressed by RB. E2F is required for chromatin synthesis. E1A-defective viruses that do not bind RB have therefore been investigated in preclinical studies with some success (McCormick 2001). Gene therapy to restore p53 function has been attempted in xenograft models bearing the leiomyosarcoma SKLMS-1 tumour cells (Scappaticci and Marina 2001). Tumour regression was achieved. Difficulty arises, however, in delivering the gene therapeutic to every tumour cell, especially in disseminated disease. The corollary to this is that delivery of the gene to normal cells must have little toxic effect (McCormick 2001).

7 The structure of the KIT receptor and the mutations associated with GISTs are discussed in greater detail in Chapter 5 on the characterisation of a new gastrointestinal stromal tumour cell line. 62

Chapter 1: Soft tissue Sarcomas

1.9.4 TMM inhibitors

The association of telomere maintenance mechanisms in tumour cells and immortality was discussed in the previous section on the biology of STS. The logical implication of this is that telomerase and ALT are potential therapeutic targets. Inhibitory strategies to date have focused on three main areas: antisense molecules (oligonucleotides, small RNA molecules) directed against hTERT, the reverse transcriptase component of telomerase, small molecule reverse transcriptase inhibitors and, more recently, small molecules capable of stabilising the four-stranded (G-quadruplex) structures formed by telomeres (White, Wright et al. 2001).

The antisense oligonucleotides comprise short stretches of DNA designed to be complementary to the target RNA (hTERT). Hybridisation to the complementary portion of hTERT inactivates it. The telomerase inhibitor GRN163L, an antisense oligonucleotide, has recently entered clinical trial phase (Burger 2007).

Dominant negative hTERT (DN-hTERT) constructs have also been successful in inhibiting telomerase both in vitro and in vivo. The introduction of DN-hTERT into bcr- abl transformed cells inhibited telomerase and reduced telomere lengths in vitro (Tauchi, Nakajima et al. 2002).

The 3a overhang of telomeres folds, forming a tetra-stranded DNA structure known as a G-quadruplex, which inhibits telomerase activity. BRACO19, a small molecule inhibitor, has been shown in a preclinical xenograft model to inhibit telomerase by stabilising these G-quadruplexes (Burger, Dai et al. 2005).

Other clinical strategies such as hTERT-directed immunotherapy (Vonderheide, Domchek et al. 2004) and siRNA technology (Kosciolek, Kalantidis et al. 2003) have also been investigated, the former as a Phase I trial with success in 4 of 7 patients and the latter, at in vitro level.

All of the above strategies target telomerase. A significant proportion of STS, as discussed earlier, utilise an alternate mechanism or ALT for telomere maintenance. Some tumours are known to possess both telomerase and ALT. In addition, the

Chapter 1: Soft tissue Sarcomas

mechanisms underlying ALT are yet to be elucidated. Whatever the mechanism, it seems clear that for any TMM inhibitor to be efficacious in STS, both telomerase and ALT must be targeted.

1.9.5 Angiogenesis inhibitors

In normal conditions, angiogenesis is usually a balanced process, involving pro- and anti-angiogenic factors. Pro-angiogenic factors include vascular endothelial growth factor (VEGF) and its receptors, bFGF and associated receptors, Angiopoietin 1 and Tie 2, matrix metalloproteases, integrins, nitric oxide and TGFβ, among others. Endogenous inhibitors of angiogenesis are thrombospondin 1, endostatin, angiostatin, and Interleukins 4, 12 and 18, to name a few (Hoekman and Pinedo 2004). In adults, angiogenesis only occurs during wound healing, inflammation, ovulation, pregnancy, or ischaemia (Scappaticci 2002). Tumours make an "angiogenic switch" in order to develop in size and metastatic potential by altering the balance of proangiogenic and antiangiogenic factors (Hanahan and Folkman 1996; Folkman 2002)8. STS tumours and cell lines are known to overexpress proangiogenic factors, such as VEGF (Chao, Al- Saleem et al. 2001; Hu, Nicolson et al. 2002; Potti, Ganti et al. 2004). This implies that selective targeting of tumour cells could potentially be carried out, with low toxicity to other normal cells.

8 The four mechanisms by which a tumour is thought to stimulate angiogenesis are discussed in greater detail in review articles on angiogenesis (Carmeliet and Jain 2000; Scappaticci 2002). The first mechanism refers to that proposed by Judah Folkman and colleagues, where tumours secrete angiogenic growth factors which may stimulate tyrosine kinase activity in endothelial cells. The second hypothesis suggests tumours can coopt existing vasculature. The third mechanism proposed is that angiogenesis may be regulated by circulating hematopoietic precursors. The fourth, is known as vascular mimicry, where tumour cells behave as endothelial cells.

Chapter 1: Soft tissue Sarcomas

Early inhibitors of VEGF such as SU5416 had promising results in preclinical xenograft models (Angelov, Salhia et al. 1999) but disappointing results were obtained from a small clinical trial (Hoekman and Pinedo 2004). A related compound, SU6668 (Laird, Vajkoczy et al. 2000) has been used in the treatment of Kaposi sarcoma. A newer compound, SU11248, designed to target VEGFR, PDGFR and KIT (all tyrosine kinases), has also been trialled (Desai, Maki et al. 2004) in patients with metastatic GISTs refractory to imatinib treatment.

Another method of targeting angiogenesis is to direct therapy against tumour endothelial cells. ZD6126 (AstraZeneca, Macclesfield, United Kingdom), a phosphate prodrug, disrupts the tubulin cytoskeleton of endothelial cells (Thorpe 2004). ZD6126 has also been used in combination with ZD6474, an inhibitor of the VEGF receptor, KDR. Other agents include TNFα (Komdeur, Hoekstra et al. 2002; Grunhagen, Brunstein et al. 2004), which is used in isolated limb perfusion (as discussed in section on chemotherapy), and is known to result in apoptosis of tumour endothelium.

1.9.6 Summary

This section has outlined the key concepts essential for an understanding of the molecular pathogenesis of soft tissue sarcomas. These have included • Differentiation of sarcomas from pluripotential mesenchymal cells, • Cytogenetic aberrations in sarcomas of simple and complex karyotype and • Escape from senescence by telomere maintenance mechanisms that involve telomerase and ALT. The relevance of these mechanisms for the development of targeted therapies has been elucidated. These agents are mostly in the early phases of pre clinical and clinical trial, with mixed results, and evidence of emerging resistance. There remains a need, therefore, to further our understanding of the molecular pathogenic mechanisms involved in these tumours and to identify further targets that can be exploited for therapeutics. The advent of high throughput techniques such as array comparative genomic hybridisation (array CGH), gene expression profiling and tissue microarray will greatly aid this process of discovery and will largely replace standard histopathologic techniques in prognostication and therapeutics.

Chapter 1: Soft tissue Sarcomas

1.10 TOWARDS a TUMOUR PROGRESSION MODEL in STS

1.10.1 Tumour Growth and Progression

Almost ten years ago, Michael Sporn wrote a commentary entitled “The War on Cancer”, reminding us of Richard Nixon’s declaration in 1971, which aimed to halve cancer related mortality by the turn of the century (Sporn 1996). Suffice to say, this goal has not been achieved. It was Sporn’s view that the goal could be better met not by adopting a reductionist approach to cancer biology but by appreciating the more fundamental concepts of tumour-stroma and tumour-host immunity interactions and importantly, the natural history of tumour progression, invasion and metastasis.

Figure 1.4 Complex Tumour-Host Interactions The reductionist approach (left panel) to cancer research considers tumours as isolated entities. Greater gains can be made by understanding fundamental concepts of tumour biology such as tumour-stroma-host interactions. In this model, adjacent “normal” cells collaborate with neoplastic cells to result in tumour progression and invasion. This figure was reproduced from “The Hallmarks of Cancer” (Hanahan and Weinberg 2000) In this vein, the hallmarks of tumour growth and progression common to all malignancies (Hanahan and Weinberg 2000) are the following • Elaboration of growth signals, • Insensitivity or resistance to growth inhibitory signals, • Escape from senescence or evasion of cell death or apoptosis,

Chapter 1: Soft tissue Sarcomas

• Ongoing replication, • Angiogenesis • Tumour invasion and metastasis. The overexpression of a number of growth factors and their receptors, such as the insulin-like growth factor (IGF) family, epidermal growth factor receptor (EGFR) and transforming growth factor beta (TGF!) has been associated with various types of cancers (Beech, Pollock et al. 1998; Busund, Ow et al. 2004; Elliott and Blobe 2005). EGFR (HER1) is a transmembrane tyrosine kinase receptor which shares homology with the other members of its family, HER2 (ErbB-2), HER3 (ErbB-3) and HER4 (ErbB-4). EGFR overexpression, which has been noted in many epithelial cancers and synovial sarcoma (Allander, Illei et al. 2002; Scaltriti and Baselga 2006), manifests as receptor activation by ligand-dependent or ligand-independent means9. The complex oncogenic role played by EGFR is becoming increasingly clear with the use of EGFR inhibitors in a number of malignancies. The mechanisms involved include EGFR gene amplification, activating mutations of the receptor or its signaling components and EGFR translocation to the nucleus, where it may function as a transcription factor for cyclin D1 (Scaltriti and Baselga 2006). Production of growth factor signals by tumour cells renders them independent of the usual homeostatic host environment. Activation of signal transduction pathways such as the SOS-Ras-Raf-MAPK pathway, ultimately cause changes in transcription and cell proliferation, among others. The dysregulation of the RB and p53 pathways which govern DNA damage repair and apoptosis is common in many sarcomas (Cance, Brennan et al. 1990; Rieske, Bartkowiak et al. 1999; Zhang, Postigo et al. 1999; Tsutsui, Kumakura et al. 2002). In conjunction with telomere maintenance mechanisms, this allows the malignant cell to evade apoptosis. Sustained angiogenesis (Hanahan and Folkman 1996; Carmeliet and Jain 2000; Folkman 2002) is another critical prerequisite for limitless tumour growth (Section 1.9).

9 EGFR signal transduction via the MAPK, PI3K/Akt and JAK/STAT pathways is considered in Chapter 4. EGFR transactivation is discussed in Chapters 4 and 8. 67

Chapter 1: Soft tissue Sarcomas

1.10.2 The Metastatic Process

Metastases account for 90% of deaths from cancer (Sporn 1996). It is thought that by the time a primary tumour is detectable, it has already undergone numerous population doublings. By the time that primary tumour is surgically resected, millions of malignant cells would have been released into the circulation. In a sense, our capacity to intervene in this seemingly inexorable course of events, still only comes into play once the horse has bolted. However, not all of these released cells will go on to develop into metastatic deposits. Only those “subclones” within the tumour cell population that have acquired the right mutations, which include the ability to migrate from the primary tumour, survive in blood or lymphatic circulation and invade distant tissues will survive to establish distant metastatic nodules.

This metastatic cascade, as summarised in Figure 1.5, initially involves invasion, by the following steps: detachment of cells from each other, attachment to extracellular matrix (ECM) and degradation of the matrix and migration. Attachment to ECM is facilitated by upregulation of laminins, fibronectin and integrins, while degradation of the ECM is carried out by enzymes such as matrix metalloproteinases (MMPs). Once intravasation has occurred, a tumour-platelet embolus forms, allowing it to “home” to its preferred site of metastasis, where extravasation occurs and the steps involved in invasion are repeated. Angiogenesis again becomes an important factor at this juncture, as the metastatic deposit establishes its vascular supply (Hanahan and Folkman 1996; Folkman 2002)10. Inherent in this concept of mutation acquisition is the notion of genomic instability. Numerous barriers to carcinogenesis are usually in place, whether repairing damaged DNA, as in the p53 pathway, or telomere shortening resulting in senescence, among others. Under normal circumstances therefore, the multiple

10 This is discussed in greater detail in the previous section on Molecular Biology and Relevance for Targeted therapy: Aniogenesis Inhibitors. Hanahan and Folkman discuss the concept of angiogenic switch and the two approaches to antiangiogenic treatment, that is, inhibition of proangiogenic factors such as VEGF and the use of endogenous inhibitors of angiogenesis, such as endostatin and angiostatin. 68

Chapter 1: Soft tissue Sarcomas

mutations required for malignant transformation, does not occur. It is of note, then, that tumour suppressor mechanisms are often inactivated in cancer, telomere lengths are maintained rather than shortening with each cell cycle, and chromosomal anomalies such as translocations occur at increased rates, resulting in fusion oncoproteins, gene amplifications and deletions. When the “caretakers” themselves become non functional, the ensuing genomic instability lends itself to the accumulation of successive mutations.

PRIMARY Transformed Cell TUMOUR

Metastatic subclone

Adhesion to / +',',1Q basement membrane IntegrinsQ ECM "' 0-,#!2',1 Host Passage through ECM lymphocyte .1Q 62� Intravasation .0-2#', 1#1

Interaction with Host immunity

Platelets Tumour cell embolus / +',',1Q ,2#%0',1Q Adhesion to "' 0-,#!2',1 basement membrane .1Q 62� Extravasation .0-2#', 1#1

Metastatic deposit .1Q ,2#%0',1Q3&"Q "&"Q:-&"Q METASTATIC TUMOUR

Venule

Figure 1.5 The Metastatic Cascade Schematic illustration of the steps involved in haematogenous spread of a tumour. Some of the factors that are upregulated to allow attachment to and invasion through ECM are shown in the ellipses. Once metastatic cells have deposited at a distant site, angiogenic factors are elaborated again to establish a vascular supply. MMP: Matrix metalloproteinase, bFGF: basic Fibroblast growth factor, VEGF: vascular endothelial growth factor, TGF!: Transforming growth factor beta. Modified from Robbins Pathologic Basis of Disease (Cotran, Kumar et al. 1994).

Chapter 1: Soft tissue Sarcomas

There are, therefore, key events that confer, in combination, the ability to escape host immunity and successfully metastasise. This metastatic phenotype must possess a molecular signature that, if present, can be detected in the primary tumour cell population and potentially, in preinvasive lesions as well. A greater understanding of the molecular basis of this process would enable us to recognise, at the time of diagnosis or surgical resection, a group of patients at greatest risk of developing metastases and therefore in need of further targeted therapy.

1.10.3 The Role of Gene Expression Profiling

1.10.3.1 In Oncology

Our current clinical system of diagnosis and management classifies tumours according to clinical presentation and histologic characteristics. Unfortunately, many related tumour types are, to a great extent, indistinguishable on conventional histologic examination, making prognostication difficult. Clinical, surgical and histologic parameters such as tumour size and resection margin may be secondary markers, while the molecular characteristics of the tumour cell population are in fact the determinants of tumour recurrence, metastasis or chemosensitivity.

In the management of cancer, the three treatment modalities of surgery, chemotherapy and radiotherapy have remained largely unchanged. Chemotherapeutic agents are generally non-selective and affect tumour as well as normal cells, resulting in toxicity. The ability to selectively target cancer cells has been exploited by newer agents such as Herceptin, a monoclonal antibody directed against ERBB2 in the treatment of breast cancer and Imatinib mesylate or Gleevec® (Novartis, Basel, Switzerland), a tyrosine kinase inhibitor of ckit transmembrane receptor in the treatment of chronic myelogenous leukaemia (CML) and gastrointestinal stromal tumours (GISTs).

Molecular analyses in the recent past have only been able to evaluate a few genes or biomarkers at a time, unable to provide an over-arching view of the genetic changes occurring in a heterogenous tumour cell population. Vogelstein’s model (Vogelstein, Fearon et al. 1988) for carcinogenesis in colorectal cancer demonstrated the “multi-hit” hypothesis applies to all cancers, in so far as multiple gene amplifications and deletions

Chapter 1: Soft tissue Sarcomas

are present in any given cancer. Moreover, as tumours progress, there is increasing genetic instability and acquisition of further mutations. This molecular complexity requires a comprehensive understanding rather than the “snapshots” we have taken in the past. What is critical for our understanding of the pathogenesis of tumours, is a more global approach – a sophisticated system of pattern recognition.

The advent of gene expression profiling (Alizadeh, Ross et al. 2001; Cheung and Spielman 2002; Holloway, van Laar et al. 2002; Miller, Long et al. 2002; Bertucci, Viens et al. 2003; Brown 2003; Prasad, Biankin et al. 2005; Oberthuer, Berthold et al. 2006; Juric, Lacayo et al. 2007) triggered an explosion of this knowledge base in oncology, enabling the screening of thousands of genes simultaneously on a single slide or chip. The major advantage of this technique is that it provides a “panoramic view” of transcriptional expression. It is the coming together of molecular biology and bioinformatics, where data on thousands of genes can be gathered from a single two day experiment. Whereas traditional gene discovery techniques are time consuming, concentrating perhaps on a single feature of a pathway, gene expression profiling provides the opportunity to grasp the complexity of interactions on a more global scale. The interplay of immunologic, inflammatory, angiogenic, developmental factors in oncogenesis, can be appreciated.

Two major microarray platforms used in this type of exploratory approach are spotted arrays, where oligonucleotide sequences are immobilised onto the glass surface, or in situ arrays such as the Affymetrix GeneChip® system (Affymetrix, Santa Clara, CA) with photolithographic synthesis. RNA is isolated from cells or fresh frozen tissue and for spotted arrays, reverse transcribed to cDNA and labelled with fluorescent dyes before hybridising to the spotted oligonucleotides on the arrays. The arrays can then be scanned to produce an image composed of varying ratios of red and green signal, providing a measure of relative gene expression.

The application of this technology in the field of cancer research has shown that we can classify and subclassify tumours on the basis of cell lineage (Khan, Simon et al. 1998; Golub, Slonim et al. 1999; Alizadeh, Eisen et al. 2000; Perou, Sorlie et al. 2000), distinguish between tumours that appear alike on histology (Khan, Wei et al. 2001),

Chapter 1: Soft tissue Sarcomas

identify the cell of origin in cases where the primary is unknown (Ramaswamy, Tamayo et al. 2001; Su, Welsh et al. 2001) and identify clinically significant prognostic groups (Alizadeh, Eisen et al. 2000; Bhattacharjee, Richards et al. 2001; Beer, Kardia et al. 2002; Bertucci, Nasser et al. 2002; Singh, Febbo et al. 2002; van 't Veer, Dai et al. 2002; Yeoh, Ross et al. 2002; Sanchez-Carbayo, Socci et al. 2003).

1.10.3.1.1 Tumour classification and cell lineage Golub et al’s seminal work provided gene profiles for two related cancers, acute myeloid and acute lymphoblastic leukaemia (AML and ALL) (Golub, Slonim et al. 1999). It was also important in so far as addressing some of the potential modes of analysing gene expression data. Golub and colleagues discussed and applied the model of “class prediction”, as distinct from “class discovery”11. Blood from 38 patients for whom the leukaemia subtype was known was used to create a predictor set of 50 genes that would distinguish the two groups. An independent group of patients with unknown leukaemia phenotypes could then be tested against this predictor set of genes and assigned to one or other group. The potential for applying this class prediction model in conjunction with clinical data such as survival and response to chemotherapy regimes became evident from this work.

Following on from this, Yeoh and colleagues12 (Yeoh, Ross et al. 2002) focused on the paediatric population with a study of 360 children with ALL. The patients were stratified into groups that would respond to, or fail therapy. Support vector machine analysis was used for class prediction, with 215 of the patients incorporated into the initial “training set”, generating a top ranked 20-gene list predictor set. The remaining 112 patients served as the “testing set”. The same supervised learning algorithm was

11 Class discovery is an unsupervised method used to define previously unknown subtypes. Class prediction involves the use of a “training set” of samples, for which the subtypes are known to define a “predictor set” of genes. This is then applied to an independent “testing set” of samples. 12 See also Table 1.8. Gene expression studies on cancer and prognosis 72

Chapter 1: Soft tissue Sarcomas

applied to compare patients in complete remission with those that relapsed and despite the heterogeneous nature of the leukaemic subtypes, a gene expression profile for predicting relapse could be generated.

The landmark study by Alizadeh et al (Alizadeh, Eisen et al. 2000) demonstrated distinct gene expression “signatures” for the single morphologic entity of diffuse large B-cell lymphoma (DLBCL), with hierarchical clustering revealing differential expression between the groups based on genes within the B cell lineage. These signatures were then shown to be associated with clinical behaviour, raising the possibility of tailored or individualised management, where one patient could be offered early bone marrow transplantation and another, chemotherapy based on his/her gene expression profile.

In 2000, another significant study (Perou, Sorlie et al. 2000) profiled 42 breast tumours and normal breast tissue, using predominantly hierarchical clustering, as developed by their group (Eisen, Spellman et al. 1998). The differentially expressed genes tended to cluster either according to their specific cell type within the heterogeneous population comprising breast tissue or according to functional pathways such as cell proliferation. The “molecular portraits” differentiated the carcinoma cells from endothelial cells, lymphocytes, macrophages and normal breast epithelium. The breast epithelium could be further distinguished by its two constituents, the basal and luminal cells. Targeted therapy for breast cancer, using tamoxifen for oestrogen-positive (ER+) cancers and trastuzumab (Herceptin®) for HER2/neu (erbB2) overexpressing ER negative cancers (Bange, Zwick et al. 2001; Romond, Perez et al. 2005) was based on earlier immunohistochemical and in situ hybridisation studies characterising breast cancers as ER+ or ERμ or those overexpressing Her2/neu (erbB2). Perou’s gene expression study identified clusters of genes associated with ER+ tumours and further stratified the ERμ group into two subtypes, one of which overexpressed erbB2.

1.10.3.1.2 Tumours with identical histologic appearance Microarray technology was applied to the diagnostic dilemma often posed by the small round blue cell tumours of childhood (Khan, Wei et al. 2001). This group of tumours, which comprise neuroblastomas, Ewing’s sarcoma, rhabdomyosarcomas (RMS) and

Chapter 1: Soft tissue Sarcomas

non Hodgkin’s lymphoma (NHL) can often be difficult to distinguish histologically due to their similar appearance. The chemotherapeutic regimes used to treat these tumours vary, hence the importance of distinguishing them. A panel of 23 tumours and 40 cell lines were examined by Khan et al. The analysis tool used was that of an artificial neural network (ANN). This is modeled on the manner in which human thought processes occur, in relation to pattern recognition, with the additional advantage of computational power. A panel of 96 discriminator genes was generated, highlights the application of class prediction models in the classification of tumours.

1.10.3.1.3 Cancer of unknown primary Other studies (Ramaswamy, Tamayo et al. 2001; Su, Welsh et al. 2001) set out to address the other common diagnostic dilemma of cancer of unknown primary. Signs and symptoms of metastatic disease can often be the initial presentation of patients. In some cases, clinical and radiologic investigation may not reveal the source of the primary tumour. The tumours at this advanced stage, can be so undifferentiated, with complete loss of architecture, that identification of cell lineage is impossible.

Su and colleagues in 2001 concentrated on the common epithelial tumours, with 100 tumour samples that included breast, prostate, lung and colorectal cancer. The class prediction model was used again and support vector machine (SVM) was employed for this purpose13. Nine of twelve metastatic tumours within the group were correctly classified for the cell lineage or primary tumour. In contrast to the previous study, a surprisingly small number of genes (eleven) could predict the origin of the test set of tumour samples with 83% accuracy. This small number of discriminators may be due to the fact that the tumours in the group had diverse cells of origin, whereas the investigations of Khan et al involved tumours thought to originate from a common

13 Support vector machine as a data analysis tool is described in greater detail by Furey and colleagues (Furey, Cristianini et al. 2000) 74

Chapter 1: Soft tissue Sarcomas

primitive cell lineage. Ramaswamy et al also employed SVM in their analysis of 218 tumours from 14 different types of epithelial malignancy and 90 normal tissue samples. The prediction had approximately 73% accuracy.

1.10.3.1.4 Cancer and prognosis A number of microarray studies have correlated gene expression data with clinical data such as survival, response to treatment, tumour recurrence and known prognostic markers (Table 1.8). While these have predominantly involved the more common epithelial malignancies, the role of microarray technology in cancer prognostication can certainly be appreciated and applied to soft tissue sarcomas.

As mentioned above, Alizadeh’s group (Alizadeh, Eisen et al. 2000) identified two subtypes of DBLCL with different prognoses. Rosenwald and colleagues (Rosenwald, Wright et al. 2002) took the analysis of the DBLCL subgroups further. They found that overall survival after anthracycline-based chemotherapy differed significantly among the subgroups, ranging from a 35% to a 60% five year survival. Cox regression analysis, identified a core set of 16 genes that were highly differentially expressed across the subgroups.

Microarray studies of breast cancer have similarly determined the prognostic value of the gene expression data (Sorlie, Perou et al. 2001; Bertucci, Nasser et al. 2002; van 't Veer, Dai et al. 2002). Sorlie and colleagues study built on their earlier work on breast cancer (Perou, Sorlie et al. 2000) and were able to subclassify the oestrogen receptor positive (ER+) group of patients further into groups with distinct clinical outcomes. Bertucci et al profiled 55 women with clinically determined poor-prognosis breast cancer treated with adjuvant anthracycline-based chemotherapy. Their findings recapitulated the work of the previous group’s, in that prognostically distinct subgroups were identified within the group of women with ER+ tumours. A difficult dilemma in breast cancer management is the development of metastatic disease in women who theoretically had good-prognosis lymph node negative disease at the time of surgical resection. Van’t Veer et al addressed this issue and identified a gene expression signature that predicted metastasis in these women. This signature comprised genes concerned with cell cycle, invasion, metastasis and angiogenesis.

Chapter 1: Soft tissue Sarcomas

Expression profiling of lung adenocarcinoma has also yielded prognostic information (Bhattacharjee, Richards et al. 2001; Beer, Kardia et al. 2002). The substratification of lung adenocarcinomas identified one group with high expression of neuroendocrine genes who had comparatively poorer survival rates (Bhattacharjee, Richards et al. 2001). A number of the predictor genes found by Beer and colleagues, such as IGFBP3, HSP-70 and Cystatin C were validated by immunohistochemistry on tissue microarrays. Similar studies in epithelial ovarian cancer (Spentzos, Levine et al. 2004) and colorectal cancer (Eschrich, Yang et al. 2005) identified gene clusters that had independent prognostic value.

1.10.3.1.5 Tumour progression and metastasis Importantly, molecular profiles of tumour progression and metastasis have also been addressed in a variety of tumour types as summarised in Table 1.9 (Okabe, Satoh et al. 2001; Ma, Salunga et al. 2003; Ramaswamy, Ross et al. 2003; Sanchez-Carbayo, Socci et al. 2003).

In a study of bladder cancer progression (Sanchez-Carbayo, Socci et al. 2003), 15 bladder tumours of varying grade were arrayed and genes examined by hierarchical clustering. The early stage tumours clustered together, and were separate from the invasive tumours. A selection of genes was studied by immunohistochemistry in bladder cancers on a tissue microarray (TMA) to confirm differential protein expression. These findings recapitulated the group’s earlier study on superficial, invasive and metastatic bladder cancer cell lines (Sanchez-Carbayo, Socci et al. 2002). Data from gene profiling of cell lines was therefore able to identify potential clinical predictors. A study of breast cancers, ranging from preinvasive in situ disease (DCIS) to invasive ductal carcinoma (IDC) posed the same question of tumour progression (Ma, Salunga et al. 2003). They identified a cluster of 29 genes differentially expressed between IDC and DCIS. Okabe and colleagues examined 10 hepatitis B positive (HBV), 10 hepatitis C positive (HCV) hepatocellular cancers (HCC) and adjacent normal liver on cDNA microarrays (Okabe, Satoh et al. 2001). Differential profiles were obtained when comparing high and low grade tumours (320 gene cluster). This cluster included genes involved in the immune system, apoptosis, cell motility and vascular invasion.

Chapter 1: Soft tissue Sarcomas

Another group was able to determine a set of genes that would predict local recurrence of disease post resection (Iizuka, Oka et al. 2003).

Singh et al similarly devised a predictor model for determining prostate cancer recurrence (Singh, Febbo et al. 2002). They were additionally able to identify gene expression signatures for the various Gleason histologic grades. 2 genes, IGFBP3 and COL1A2 correlated with both grade and patient outcome.

Obtaining a “metastasis signature” for adenocarcinomas of various types was the goal of one study involving 64 primary tumours and 12 metastases (Ramaswamy, Ross et al. 2003). A 128-gene cluster was developed, and applied to available data on a further 279 primary solid tumours. The gene signature correlated with metastases and poor outcome in these cases, indicating that for epithelial tumours, there exists a metastatic gene profile that is present in the primary tumours that go on to metastasise.

Chapter 1: Soft tissue Sarcomas

Table 1.8 Gene expression studies on cancer and prognosis in Epithelial Malignancies S12832&-0 D-30, *?0 +.*#1 '!0- 00 7 33+-30.0-%,-1'1 8*'8 "#&8 < 230#RR [X,-0+ * ," SYZWX! <8/7+.&-!&'. /*/13 27.#15'2&"'$$#0#,2.0-%,-1#1 + *'%, ,2*7+.&-+ 1 +.*#1 (-1#,5 *"8 <-DRT TVR /*/ STS[X! <8/7+.&-!&'. 304'4 *.-12!&#+-2� .7 /*/ ?#-&-D * *#**RT UXR. #"' 20'!8// 8$$7+#20'6>[W84T . #"' 20'!8//S.0#"'!2'-,-$0#* .1#41!-+.*#2#0#+'11'-, -0*'#3 .<8[Z YZ 0# 12! ,!#01 ZSRT! <8 00 7 -(&0# 12* ,!#0.0-%,-12'!13 %0-3.1 : ,2:##0/D < 230#RT [Z! ,!#01R TWRRR! <8 00 7 0# 12* ,!#0R*-00#* 2'-,2-"#4#*-.+#,2-$+#2 12 1#1 1-+#(*8S ,"T 1-+#+#215'2&',W701 #023!!'" %3+-*&#,RT WW.--0.0-%0# 12* SRVW! <8 00 7 0# 12* ,!#01304'4 *R-(&.0-%,-12'!13 %0-3.1 20# 2#"5'2&!&#+-(6 ##0 & < 2#"RT ZX*3,% "#,-! 0!',-+ -*'%-,3!*#-2'"# 00 7 /3,% "#,-! 0!',-+ 1 SR,-,,#-.* 12'!1 +.*#1 & 22 !& 0(##8 .<8RS SZX*3,%23+-301 STXRR-*'%-,3!*#-2'"# 00 7 /3,%! ,!#0R.--0.0-%,-1'1^,#30-#,"-!0',#_13 %0-3. ( + 15 +7 < 2&#,RU TY[.0'+ 0723+-301 : 0'-318$$7+#20'6 00 71# -*'"23+-301+#2 12 2'!1'%, 230# '83) < / ,!#2RU UU%** XRRR-*'%-,3!*#-2'"# 00 7 %**.0#"'!2'-,-$0#!300#,!# .#,28-1 D*6′RV XZ #.'2&#*' *-4 0' ,! ,!#016*'%-,3!*#-2'"# 00 7 SSW%#,#1'%, 230#-$',"#.#,"#,2.0-%,-12'!4 *3# -1!&0'!& D*6′RW YZ*(* UT7! <8 00 7 VU%#,#1'%, 230#.0#"'!2',%-4#0 **1304'4 * /*/S '$$31#* 0%#V!#***7+.&-+ Q8//S8!32#*7+.&- * 12'!*#3) #+' Q*(*S*-*-0#!2 *! ,!#0Q-(&S6#120-%#,0#!#.2-0.-1'2'4#23+-301Q%**S%#. 2-!#**3* 0! ,!#0Q /S/#'-+7-1 0!-+ Q%#.&S%#. 2'2'1.-1'2'4#Q%#.*&S%#. 2'2'1*.-1'2'4#Q#21S#2 12 1#1Q *S 3!2 *! 0!',-+ ',1'23Q8 %S827.'! *"3!2 *&7.#0.* 1' Q *S,4 1'4#"3!2 *! 0!',-+ Q

Chapter 1: Soft tissue Sarcomas

Table 1.9 Gene Expression Studies addressing Tumour Progression "32&-01F-30, * +.*#1"00 71',"',%1 ,!V* 0 7-#2 *8D.′RU V* ""#0! ,!#0!#***',#1Q SYZVT! <8 [T%#,#!*312#0"'$$#0#,2' 2',%',1'23 SW23+-301*-5 ,"&'%&%0 "# ,"',4 1'4#23+-301R304'4 *!-00#* 2'-, 6) #%#2 * * (#1′RS TR%** ," "( !#,2,-0+ *2'113#R TURVR! <8 UTR%#,#!*312#0"'12',%3'1&',%&'%& ,"*-5%0 "#1R SR%#.& ,"SR%#.*& '$$#0#,2' *.0-$'*#1$-0%**4#0131,-0+ * ," %#.&4#0131%#.*& HD#2 * .<8′RU UX0# 12! ,!#0S.0#',4 1'4#Q ! <8 '$$#0#,2' *#6.0#11'-,$-0 *4#0131 *R 8 %Q *Q *RU<-0+ *8 %!*312#015'2&*-5%0 "# * /##?"#2 * * (#1′RV UY/STR.0'+ 07QYQ WXRU! <8 ZR%#,#!*312#0"'12',%3'1&',%.0'+ 074#0131 V.QX/(+#2 12 1'1',!*3"',%%#,#1 XQ WQ X ( + 15 +7#2 *< 2&#,′RU XV.0'+ 07 "#,-! 0!',-+ 1Q 8$$7+#20'6 00 71# STZ%#,#!*312#0$-0+#2 12 1'1b1'%, 230#c ST#2 12 1#1 ',%& #2 * * *#**′RT WT.0-12 2#! ,!#01RWR<-0+ *.0-12 2# 8$$7+#20'6>[W84T <-%#,#!*312#0 11-!' 2#"5'2&*-! *',4 1'-,R W%#,#+-"#*2-.0#"'!20#!300#,!# ,"T[%#,# !*312#0"'12',%3'1&',%.0-12 2#! ,!#01-,&*# 1-, 1!-0# %**S%#. 2-!#**3* 0! ,!#0Q/S/#'-+7-1 0!-+ Q%#.&S%#. 2'2'1.-1'2'4#Q%#.*&S%#. 2'2'1*.-1'2'4#QS#2 12 1#1Q.S.0'+ 0723+-305'2&!-,!300#,2+#2 12 1#1Q /(S/-! *0#!300#,!#Q *S 3!2 *! 0!',-+ ',1'23Q8 %S827.'! *"3!2 *&7.#0.* 1' Q *S,4 1'4#"3!2 *! 0!',-+ Q

Chapter 1: Soft tissue Sarcomas

Table 1.10 Gene Expression Studies on Soft tissue Sarcomas S12"32&-0 F-30, * +.*#1 #'!0- 00 71 *312#01#,#1-$ ,2#0#12 7& ,D * (#1[Z Y8(!#***',#1 STUZ! <8 UY%#,#!*312#0$-08(!$T0#$#0#,!#!#***',# 7& ,D .<8[[ .8HU ,".8HUV"7%(20 ,1$#!2#"<%U3U!#***',#1 TTTW! <8 .8HUV"7%(*',#S/ [Q*3%Q4W 8** ,"#0: * (#1RS SU&31X1.',"*#!#**23+-301 SUQZTX! <8 SSU%#,#!*312#0&31S #$@Q #XVQ2/#/\Q < % 7 + * (#1RT SURUV-2 TURVR! <8 .<3!*312#0R"%Q/Q /Q./"-,-2 <'#*1#,36 / ,!#2RT Z&31RSS/RZRZ"%RV/.RT.<3 TT7QVT7! <8 &3!*312#0R3 %0-3./!*312#0 &+3*#4'!& * ,!#0RT U/R 2!&#"< TURV! <8 ','+ *',20 V23+-304 0' '*'27 8** ,"#0: 8D.RT SVQV"%Q ,"S" XWVU! <8 SWU%#,#!*312#0$-0S/ XQ/XQ#X !& #$#07/ :'0!&80!&RT &&VXT.#62+ **0-3," *3#!#**23+-30*',#1 8$$7&SSR *#***',#"'",-2!*312#05'2&-5',%1 F ' % ,2D6,!RT SS!#***',#1S-5',%1Q<Q. 8$$7&SSR <S/8XQ $T-5',%1S Q2 Q$$2WQ2X (#, %3+. 2&RU UW3SSS/-$ **%0 "#1 8$$7>SUU8 /%#,#!*312#0S13 %0-3.1$-023+-30"'$$#0#,2' 2'-, )3 '287 * ,!#0RU V>2#0',#/QV<-,>2#0',#/QS[/#'-+7-+ 1QVX< 8$$7>[W8 VTW%#,#11V$-*""'$$#0#,2' **7#6.0#11#" #% *< 8D.RU SS"%QZ"QX/QU./QW&31QWQV**3#QW /. 8$$7>[W8 Q /.Q"!*312#01T (-%0#,% 8D.RU T-$SR-*1 VRRR! <8 UW%#,#!*312#0S$XQ2/YQ/ Q X /##?" D*RU TY3 WXRU! <8 %#,#!*312#0R"%/,-"'12',!2#6.0#11'-,.0-$'*# /##?" * (#1RV UY/RTR.0'+ 0723+-301Y#21 WXRU! <8 ZR%#,#!*312#0$-0+#21S WQ XQ X !& #$#07/ * (#1RV X2STRTT**3#!#***',#1RTS-2�*',#1 8$$7>[W84T URR%#,#!*312#0**3#S#YQWVV WWQ+0VX '0"7 * (#1′RW SZS -,# ,"3 STXRS! <8 **312#01$-0&3Q/XQQ6Q(Q ".Q-5',%_1 !& #$#07/ <#-.* 1' ′RX Z**3QST-2�. #"' 20'!1 0!-+ !#***',#1 *&% 00 71 2#WQ#Y ?--, D30%(#1′RX UZ3 6*'%- 00 71 8,%'-%#,#1'1!*312#0S /# Q XQ21$Z "0 ,!'1. *&#,′RY SYY3 TY7! <8 &3 76-'"/.!*312#0S/#Q//#QF,2Q<-2!& 3&# -4# '1 1#*#!2#"Q ,-2 #6& 312'4# *'12Q -, %#,# #6.0#11'-, 123"'#1 -, 3T 8$$7S 8$$7+#20'6Q 8(S 8*4#-* 0(& "-+7-1 0!-+ Q**3S**# 0!#**1 0!-+ -$1-$22'113#Q*&%S !-+. 0 2'4#%#,-+'!&7 0'"'1 2'-,Q /.S #"'$$#0#,2' 2#"/'.-1 0!-+ Q ".S #0+ 2-$' 0-1 0!-+ .0-203 #0 ,1Q-*S-620 1)#*#2 *+76-'"!&-,"0-1 0!-+ Q"S"' 0-1 0!-+ Q &3S & 120-',2#12', * 120-+ * 23+-30Q /S /#'-+7-1 0!-+ Q "%S *'%, ,2 "' 0-31 %'12'-!72-+ Q .S *'%, ,2 +#* ,-+ 1-$2 . 021Q

~~Chapter 1: Soft tissue Sarcomas~~

~~1.10.3.2 In Soft tissue sarcoma~~

One of the main problems in the management of STS is its diverse nature with respect to both tumour location and histology, with over 50 documented histologic subtypes of STS. In addition, the histologic classification of certain subtypes continues to be debated and revised. The histologic subtype is thought to influence the clinical behaviour of the malignancy, hence the importance placed on classification. The majority of gene expression studies in STS to date, have therefore addressed the issue of histologic class (Table 1.10).

1.10.3.2.1 Paediatric-type Sarcomas A number of studies utilised a panel of cell lines to delineate gene clusters. These have typically been sarcomas that are more common in the paediatric population such as Ewing’s sarcoma (ES), alveolar rhabdomysarcoma (ARM) and the histologically similar tumours grouped together as the small round blue cell tumours (Khan, Simon et al. 1998; Schaefer, Wai et al. 2002; Wai, Schaefer et al. 2002; Schaefer, Brachwitz et al. 2004; Schaefer, Brachwitz et al. 2006). In contrast to the most common adult STS such as leiomyosarcomas (LMS) and liposarcomas (LPS), reasonable numbers of cell lines exist for the aforementioned tumours.

Wai and colleagues (Wai, Schaefer et al. 2002) examined the differential profiles of Ewing’s sarcoma, neuroblastoma and malignant melanoma soft parts (MMSP) cell lines. From the same group of researchers, Schaefer et al identified ERBB3 as one of the most dramatically up-regulated genes in clear cell sarcoma. Their earlier study (Schaefer, Wai et al. 2002) had used microarray to reclassify a cell line GG-62, thought to be derived from a Ewing sarcoma. From hierarchical clustering, the most recent study from this group identified a further gene, neuregulin (NRG1) as upregulated in the clear cell sarcoma of soft tissue (CCSST) cell lines (Schaefer, Brachwitz et al. 2006). NRG1 is a ligand in the EGF family.

Khan et al, in two of the earliest publications on microarrays in sarcoma, examined the profile of ARM (Khan, Simon et al. 1998) using 7 ARM cell lines, then went on to perform a functional study, in which NIH 3T3 cells were transfected with the PAX3-

~~Chapter 1: Soft tissue Sarcomas~~

FKHR fusion oncoprotein that characterises ARM (Khan, Bittner et al. 1999). The aim was to identify the genes that were activated or inhibited downstream of PAX3- FKHR14. The fusion gene was shown to induce a number of genes involved in muscle transcription.

1.10.3.2.2 Tumour samples for histologic classification Microarray based subclassification of STS has produced sets of gene predictors for the sarcomas of relatively simple karyotype, such as GISTs, myxoid liposarcoma, Ewing’s and synovial sarcoma (SS) (Khan, Simon et al. 1998; Allander, Nupponen et al. 2001; Allander, Illei et al. 2002; Nielsen, West et al. 2002; Lee, John et al. 2003; Baird, Davis et al. 2005). Allander and colleagues showed SS to cluster separately from malignant fibrous histiocytoma (MFH) and Fibrosarcoma.SS, in particular expressed higher levels of IGF2 but had lower expression of IGFBP2. Gene expression differences between the monophasic and biphasic forms of SS were also discovered. Nielsen and colleagues found that the sarcomas of complex karyotype such as the leiomyosarcomas, liposarcomas and malignant fibrous histiocytomas, had less distinct clusters, although 6 of 11 LMS did form a cluster, overexpressing genes involved in muscle structure such as actin and myosin. Their SS cluster contained genes such as EGFR, TGFB2, SSX4 and SSX3. There were differences, however, between the gene clusters obtained for SS in the 2 studies (Allander, Illei et al. 2002; Nielsen, West et al. 2002). The two studies had a number of methodological points of difference such as the choice of reference cell lines (one osteosarcoma cell line versus pooled mRNA from 11 cell lines), the platforms (13K arrays in the former, versus a combination of 22K and 42K cDNA arrays in the latter) and the analysis methods used (single value decomposition in the latter study to reduce bias caused by the use of two array types).

~~14 The PAX3-FKHR fusion gene has been discussed in the previous section Biology of STS: Sarcomas of simple karyotype (Section 1.8.3.1). 82~~

~~Chapter 1: Soft tissue Sarcomas~~

These differences may in part explain some of the differences in results seen in the two studies. More recent reports of gene expression arrays in STS have corroborated this tight clustering of synovial sarcomas and other sarcomas of simple karyotype such as dermatofibrosarcoma protruberans (DFSP) and GISTs (Baird, Davis et al. 2005; Francis, Namlos et al. 2007).

The GIST cluster outlined by Nielson et al included KIT, Protein kinase C, sprouty homologues 1 and 4 (SPRY1 and SPRY4). The study showed GISTs and LMS, which used to traditionally be grouped together, as two distinct entities in terms of their gene expression profiles.

Malignant peripheral nerve sheath tumours (MPNST), as would be expected, have been shown to overexpress nerve sheath related genes (Nagayama, Katagiri et al. 2002) and dedifferentiated liposarcomas (DDLPS) overexpressed CDK4 and MDM2, while melanocytic related genes such as SOX10, gp100 and MITF (microphthalmia-associated transcription factor) characterised the expression profile of clear cell sarcoma, also known as malignant melanoma soft parts (MMSP) (Segal, Pavlidis et al. 2003).

Comparative genomic hybridisation (CGH)15 and cDNA microarray were both performed on 16 liposarcomas (Fritz, Schubert et al. 2002). Dedifferentiated liposarcoma (DDLPS) showed amplification of MDM2, GLI and CDK4, all on chromosome 12q13-15.

A small but methodologically important study was carried out on three leiomyosarcomas (LMS) to address issue of tumour heterogeneity in these large tumours (Shmulevich, Hunt et al. 2002). Tissue was obtained from central and

~~15 In array CGH, genomic DNA rather than RNA is used to perform a whole genome analysis, chromosome by chromosome, analysing large scale chromosomal copy number gains. 83~~

~~Chapter 1: Soft tissue Sarcomas~~

peripheral regions of the same tumour. Using multidimensional scaling and hierarchical clustering, peripheral and central samples from the same tumour clustered together. Intra tumour heterogeneity was in other words insignificant, indicating that the sampling of tumours for gene expression analysis could be carried out from a single region of a large tumour.

Lee at al studied 51 STS of which 9 were LMS (Lee, John et al. 2003). The study did not find a specific cluster for LMS, rather they tended to group together with malignant fibrous histiocytomas. The synovial sarcomas did cluster. The gene cluster obtained in this study for SS did not, however, correlate well with the set identified by Nielsen and colleagues, discussed above. Of note, EGFR was not present in the SS cluster defined by Lee and colleagues, in contrast to two of the other studies of SS (Allander, Illei et al. 2002; Nielsen, West et al. 2002) but in agreement with a third study (Nagayama, Katagiri et al. 2002). This discrepancy has therapeutic implications and validation of the data on different sets of SS is required.

1.10.3.2.3 STS, Microarrays and Tumour Progression With the exception of a few STS, such as the paediatric sarcomas, the majority of STS, whatever their histologic subtype, remain insensitive to traditional chemotherapeutic agents. The need to develop novel therapeutic agents is therefore acute. The success of Imatinib in the treatment of unresectable gastrointestinal stromal tumours (GISTs)16 has provided the impetus for research in this field. Predicting which patients presenting with primary disease will develop metastases is the key to this individualised approach. Microarray technology needs to be applied to identifying the gene signatures associated with disease progression and metastasis in STS.

~~16 Targeted therapy with Imatinib was introduced in the preceding section on Molecular Biology and Relevance for Targeted Therapy: Oncoproteins: Transcription Factors and Tyrosine kinases 84~~

~~Chapter 1: Soft tissue Sarcomas~~

Three studies to date, have examined disease progression or metastasis and outcome, two of these concentrating on LMS (Ren, Yu et al. 2003; Lee, John et al. 2004). The earlier study comprised 35 tumours, 11 of which were leiomyosarcomas and the later study, 37 LMS, 20 of which were primary tumours and 7 were metastases. The third and most recent study, examined a subset of 89 STS that were categorised as MFH or undifferentiated pleomorphic sarcomas, 76% of which were high grade tumours (Francis, Namlos et al. 2007).

The first study found LMS gene expression patterns to correlate with tumour differentiation. An “index of malignancy” was devised to express the differentiation and metastatic potential of the tumour and this was used as the measure of clinical aggressiveness, rather than survival data represented as Kaplan-Meier plots. The more recent study on LMS to examine the concept of predicting metastases (Lee, John et al. 2004) did actually include metastatic tumour samples. Supervised class comparison was used to obtain a list of 335 differentially expressed genes, the “metastatic signature”, with which further analyses such as supervised learning methods were carried out. The supervised learning narrowed the list of discriminator genes to 80. The Kaplan-Meier plots for the 2 groups of patients showed distinct survival differences.

Francis et al carried out a supervised analysis of 89 pleomorphic sarcomas, generating a 200-gene list with a high false discovery rate (FDR) of 35% for discriminating tumours that would or would not metastasise (Francis, Namlos et al. 2007). Their leave-one-out cross validation analysis assigned 64% of their samples to the correct low and high risk groups.

These three studies are an indication that metastasis and clinical outcome in mesenchymal tumours can be explored using global gene expression arrays, providing us with much needed insights into the molecular pathogenesis of STS.

~~1.10.4 Validation of Results in Gene Expression Studies~~

Gene expression arrays are capable of efficiently generating a large amount of data in a short period of time. However, there are as yet no, or few, gold standards, in terms of methodology. Significant sources of variation exist from collection of the tissue and

~~Chapter 1: Soft tissue Sarcomas~~

~~extraction of the RNA, through to experimental design and analysis tools. These methodological issues will be discussed further in Chapter 2.~~

For these reasons, gene expression analyses can be considered to be sophisticated “hypotheses generators”, requiring further validation. This can take the form of the traditional “gold standard” method for detecting mRNA expression, Northern analysis, RT-PCR or real time RT-PCR (qRT-PCR) are alternative methods that allow for higher throughput. mRNA expression may not necessarily correlate with protein expression for a variety of reasons, including splice variants, post-transcriptional and post-translational modification. The development of tissue microarray technology has paved the way for the analysis of potentially hundreds of specimens using immunohistochemistry or fluorescence in situ hybridisation. The application of both qRT-PCR and tissue microarrays for further evaluation of candidate genes in sarcoma tumour progression will be discussed in greater depth in Chapters 3 and 4.

~~1.10.5 The State of the Science~~

Soft tissue sarcomas, as discussed, comprise a heterogeneous group of tumours of some 50 different histologic subtypes. However, for the ultimate purpose of identifying novel therapeutic targets, the salient unifying features to be considered are these: • Tumour grade remains the strongest independent clinicopathologic predictor of metastasis and mortality • Key factors are involved in the lineage specific differentiation of STS from a common pluripotent mesenchymal precursor • Underpinning the diverse cytogenetic aberrations seen in the more common STS subtypes is genomic instability • A more global molecular approach to sarcoma research, rather than examination of individual histopathologic or clinical factors, will elucidate the patterns and functional pathways involved • Gene expression profiling has enabled molecular classification of STS but the gene signatures associated with tumour grade and progression of disease are yet to be fully explored.

~~Chapter 1: Soft tissue Sarcomas~~

The problems we are faced with in this field are that • 50% of patients die from metastatic disease, hence the urgent need to identify potential therapeutic targets to treat this at-risk group of patients. • There remains a paucity of sarcoma cell lines that can be used as models for preclinical drug trials • The relative rarity of sarcomas compared to epithelial tumours often restricts the sample size of any single institution study. What is required, therefore, is the development of new sarcoma cell lines that can be added to the repository, multi-institutional tumour banks of frozen and paraffin- embedded tissue and finally, the identification of new molecular markers or functional groups of genes that determine tumour progression and metastasis, with the ultimate aim of developing targeted therapeutics.

~~1.10.6 The Aims of this Thesis~~

In light of the above issues, this study aims to • Examine global gene expression patterns associated with tumour progression and metastasis by gene expression profiling of three sarcoma cell lines of increasing metastatic potential, • Select significant genes such as EGFR for further study and validate their expression at transcript level using quantitative real time reverse transcriptase polymerase chain reaction (qRT-PCR), • Determine the predictive value of EGFR and its signal transducers in sarcoma tumour progression and survival and finally, • To concurrently establish and characterise new sarcoma cell lines that can, in the future, serve as in vitro models for functional studies on EGFR and other markers.

~~Chapter 2: Gene Expression Patterns in STS Progression~~

~~2",~~

~~2. IDENTIFICATION of GENES and PATHWAYS in SOFT TISSUE SARCOMA PROGRESSION by GENE EXPRESSION PROFILING~~

~~Chapter 2: Gene Expression Patterns in STS Progression~~

~~2.1 INTRODUCTION~~

The majority of patients with soft tissue sarcoma present with high grade tumours. These are the patients at greatest risk of developing metastatic disease, to which 50% of them will succumb. Conventional chemotherapeutic agents have little success. There is a pressing need to identify genes and pathways involved in tumour progression in these patients, in order to select molecular markers of prognostic and therapeutic significance. Previous studies have proven the ability of expression arrays to discriminate between histologically indistinguishable tumours and identify clinical and prognostic subclasses of tumours. However, few have directly addressed the question of tumour progression and metastasis in all histologic subtypes of soft tissue sarcoma (STS).

This section describes the methods and results obtained in the gene expression profiling of sarcoma cell lines of increasing metastatic potential, with the aim of generating hypotheses for tumour progression and metastases in soft tissue sarcomas. Prior to outlining the methods, however, certain concepts regarding study design warrant introduction.

~~2.1.1 Design Issues in Expression Profiling~~

Gene expression profiling is a technique that has now been used within the research community and the importance of experimental design is well accepted (Churchill 2002; Dobbin and Simon 2002; Yang and Speed 2002; Dobbin, Shih et al. 2003; Holzman and Kolker 2004). There are both scientific and logistic issues that govern study design. The former deals with ascertaining whether the design of the experiment will answer the questions posed. The latter is concerned with the practical aspects such as availability of sample material, choice of reference RNA and standardisation of all protocols to control for experimental variability.

~~2.1.1.1 Replicates~~

~~Chapter 2: Gene Expression Patterns in STS Progression~~

~~hybridisations carried out using RNA from the same extraction. Replication is essential in experimental design because it allows an estimate of variability.~~

~~2.1.1.2 Experimental Design~~

The design of the experiment may be direct, where two samples of interest are compared, or indirect, where many samples are compared to a common reference (Dobbin and Simon 2002; Yang and Speed 2002). A variant of the direct design is the loop design (Kerr and Churchill 2001a; Kerr 2003), where samples are successively compared. The inherent problems with the loop designs are the complexity of the statistical analysis that has to be performed and difficulty in determining the accuracy of the comparisons if the samples are far apart on a large loop.

~~2.1.1.3 Dye Bias and Other Sources of Variation~~

As dual colour arrays require the hybridisation of red and green labelled cDNA samples with the oligonucleotide probes on the slides, the ratio of the green to red fluorescence signal is what is obtained on scanning. This is therefore a measure of the relative expression levels of the transcripts. As gene specific dye bias is known to occur, as well as differences in the dye incorporation efficiencies across all genes, these may confound the relative gene expression patterns observed. Normalisation (see below) will correct for the dye incorporation differences but not individual gene-specific dye binding differences. This has led to some researchers performing dye swap experiments to control for this confounder. However, a common reference design can be used without dye swaps where the comparison of interest is between samples that are all labelled with the same fluorescent dye. Differences between arrays may arise from differences in print quality. Other sources include everything from differences in RNA quality to post hybridisation stringency washes. The effects of these variations can be large relative to the actual biological differences between the samples.

~~2.1.1.4 Normalisation and Analysis of Data~~

2.1.1.4.1 Normalisation The purpose of normalisation is to remove systemic sources of variation such that the real biologic differences between samples can be measured (Quackenbush 2002; Yang, Dudoit et al. 2002; Bolstad, Irizarry et al. 2003; Smyth and Speed 2003). Normalisation

~~Chapter 2: Gene Expression Patterns in STS Progression~~

needs to be intensity dependent and carried out between as well as within arrays. Global per spot per chip intensity dependent (LOWESS)17 normalisation the most commonly used method for achieving this. The mathematical models are described by Yang and colleagues and in a review by Quackenbush (Quackenbush 2002; Yang, Dudoit et al. 2002).

2.1.1.4.2 Data Analysis Numerous methods have been employed in microarray data analysis. Most early studies used “fold change” above a user defined threshold to present the data. Current practice applies a variety of standard statistical tools modified for use in this setting where large datasets require analysis. This includes parametric tests such as the t-test and analysis of variance, or ANOVA18 (Kerr, Martin et al. 2000; Slonim 2002). Multiple testing correction is applied to reduce the false discovery rate (FDR). Standard methods such as Bonferroni, Westfall and Young permutation or Benjamini Hochberg (decreasing order of stringency) can be used for this purpose19. Higher order analyses such as clustering and class prediction can be carried out with the reduced list of statististically significant genes.

~~2.2 MATERIALS AND METHODS~~

~~2.2.1 Cell Lines~~

~~All cell lines used in this part of the study were purchased from American Type Culture Collection (ATCC) and cultured in Dulbecco’s Modified Eagle’s Medium (DMEM)~~

17 LOWESS refers to Localised Weighted Regression. This was originally described in Cleveland, W. S. and S. J. Devlin (1988). Locally-weighted regression: An approach to regression analysis by local fitting. J. Am. Stat. Assoc. 83, 596–610. 18 ANOVA is used to assess the variability, or spread of data. Variance is the average squared deviation from the mean. It measures the spread of data around the mean. 19 The listed methods for multiple testing correction can be applied to the ANOVA within the GeneSpring analysis program used for the microarray analysis in this study, as discussed in Section 2.2.8.1 under Materials and Methods. The Benjamini Hochberg false discovery rate was used in this case (Benjamini and Hochberg 1995). 91

~~Chapter 2: Gene Expression Patterns in STS Progression~~

with 10% foetal calf serum (FCS), penicillin (100U/ml) and streptomycin (100U/ml). MRC5, a normal lung fibroblast cell line was used as the common reference. Three sarcoma cell lines of increasing metastatic potential were used (Masters and Palsson 1999): SW684, a low grade fibrosarcoma (FS) (Fogh, Fogh et al. 1977), HT1080, a well characterised high grade FS (Rasheed, Nelson-Rees et al. 1974; Timar and Paterson 1990; Slovak, Coccia et al. 1991; Slovak, Ho et al. 1993; Trautinger, Kokesch et al. 1999; Kim, Jung et al. 2004; Sloan, Eustace et al. 2004) and GCT, as the metastatic cell line. GCT is a cell line derived from a malignant fibrous histiocytoma (MFH) lung metastasis (Di Persio, Brennan et al. 1978; Hsu, Rohol et al. 1989; Liesveld, Rush et al. 1993)20.

Cell proliferation rates were determined using the Dimethylthiazoylyldiphenyltetrazolium bromide (MTT) assay (Sigma-Aldrich, Castle Hill, Sydney, NSW) (Appendix I). MTT assay is a colorimetric assay (an assay which measures changes in color) which can be used to indirectly measure cellular growth (Mosmann 1983; Plumb 2004). Yellow MTT (3-(4,5-Dimethylthiazol-2-yl)-2,5- diphenyltetrazolium bromide, a tetrazole) is reduced to purple formazan in the mitochondria of living cells. A solubilisation solution, in this case dimethyl sulfoxide is added to dissolve the insoluble purple formazan product into a coloured solution. The absorbance of this coloured solution can be quantified by measuring at a certain wavelength by a spectrophotometer. This reduction takes place only when mitochondrial reductase enzymes are active, and therefore conversion can be directly related to the number of viable (living) cells. The proliferation or cellular growth rate of the cells (in hours) can then be derived from a graph which plots absorbance against hours of cell culture, using the portion of the graph where the absorbance is exhibiting exponential change.

20 The theories on the cell of origin of MFH were discussed in Chapter One (Section 1.4.1). One of these is that the cell of origin is the fibroblast (Antonescu, Erlandson et al. 2000; Suh, Ordonez et al. 2000; Erlandson and Antonescu 2004), based on the ultrastructural similarity of MFH to fibrosarcomas. 92

~~Chapter 2: Gene Expression Patterns in STS Progression~~

~~2.2.2 19K Arrays and Quality Control of Printing~~

These arrays comprised a commercial oligonucleotide library from Compugen (Compugen 19,000 human oligo library Release 1) and manufactured by Sigma- Genosys. Information on the 19K oligo library can be found at http://www.ramaciotti.unsw.edu.au/groups/microarray/data/human/. The library consists of 18 861 oligonucleotides and 197 controls. Additional controls to monitor printing and non specific binding are also printed on the arrays (Appendix I, Table I.1). Quality control (QC) of the printing was carried out by the respective facilities, representative examples of which are included in Appendix 1, Figure I.1. Slides taken from the beginning, middle and end of the print run were used for this purpose. Normalisation of the data was carried out. Individual features or “spot morphology” was also taken into consideration, such as the presence of “donuts” (halos of concentric red and green signal) or “comets” or the presence of signal in the positions that contained no oligonucleotide probes, just buffer. These may indicate problems with the print tips or contamination of the buffers.

~~2.2.3 Experimental Design~~

A common reference design was chosen (Kerr and Churchill 2001b; Churchill 2002; Kerr 2003; Holzman and Kolker 2004) (Figure 2.1), using pooled MRC5 as the reference, labelled on all slides with the green fluorescent dye, Alexa Fluor®555 (Invitrogen, Carlsbad, CA). The common reference design avoids the problem of unequal dye incorporation efficiencies, by always comparing the samples labelled with red dye (Alexa Fluor®647). As the tumour cell lines are always labelled with Alexa Fluor®647 and no comparison is being made directly to the Alexa Fluor®555 labelled MRC5, the need for additional dye swap experiments is obviated. For each cell line, at least triplicate arrays were performed. Cell culture conditions were standardised for all cell lines and RNA extractions were carried out on different days at 70-80 % confluency from the cell lines at different passages. These formed the biological replicates.

~~2.2.4 RNA Isolation~~

~~To limit experimental variability, all cells were harvested at 70-80 % confluence, that is, 70-80% of the area of the culture dish was covered by cells at the point of harvest. As~~

~~Chapter 2: Gene Expression Patterns in STS Progression~~

the different cell lines had slightly different rates of proliferation, the time taken to reach 80% confluence differed for the cell lines. While we understand that a cell count should ideally be carried out for each harvest, as well as judging it by eye, this is a routine method used in cell culture and consistency is known to be achieved. Performing a cell count would also potentially detract from the ability to extract good quality RNA, which requires that the processing or snap freezing be carried out with care and speed so as to prevent degradation or contamination. In this study, TRIzol® (GibcoBRL, Invitrogen, Carlsbad, CA) was used to extract RNA. This was added directly to the culture dishes, and the cells scraped in TRIzol® then immediately snap frozen for later RNA extraction. This yielded excellent quality RNA as determined by subsequent analysis using both the Agilent 2100 BioAnalyzer and NanoDrop ND-1000. RNA was extracted for both microarray and qRT-PCR experiments. RNA from MRC5 at various passages was extracted and pooled before aliquoting, to ensure uniformity of the reference sample on each microarray slide. Pooling aimed to control for any variation in the reference sample. The detailed protocol is listed in Appendix I.

~~2.2.5 Assessment of RNA Quality~~

~~2.2.5.1 Agilent 2100 BioAnalyzer for RNA Integrity~~

High quality RNA is crucial for the success of microarray experiments. The quality and quantity of the RNA was confirmed on the Agilent 2100 BioAnalyzer (Agilent Technologies, Palo Alto, CA) (Genter, Burman et al. 2002; Rainen, Oelmueller et al. 2002; Thelen, Burfeind et al. 2004) using the RNA 6000 Nano LabChip kit. The BioAnalyzer uses capillary gel electrophoresis with an intercalating dye to provide a gel image and electropherograms, together with the 28S/18S ratios and concentrations. The integrity of the RNA was assessed for degradation and contamination in this manner. The BioAnalyzer is accurate for a concentration range between 50-500 ng/μl. Samples were therefore diluted before loading 1 μl of each sample onto the 12 sample chip (2-6 % SD from UV readings).

~~2.2.5.2 NanoDrop ND-1000 for RNA Quality and Quantity~~

~~The RNA concentrations and 260/280 ratios were also confirmed on the NanoDrop ND- 1000 (NanoDrop Technologies, Delaware, USA). This is a spectrophotometer that uses~~

~~Chapter 2: Gene Expression Patterns in STS Progression~~

~~ 3+-30#***',#1~~

~~,XZV -50 "#~~

~~#W 2SRZR '%&0 "# (-0+ * ' 0- * 12 #2 12 2'! ~~

~~"*#6 WWW "*#6 XVY ~~

Figure 2.1 Experimental Design for Gene Profiling of Sarcoma Cell Lines RNA was isolated from the cell lines and reverse transcribed to cDNA. A common reference design was used (a), where the normal fibroblast cDNA (MRC5) labelled with green fluorescent dye and one of the red labelled tumour cell lines was hybridised onto a glass slide containing the 19000 genes. These slides were scanned on a GenePix 4000B laser scanner. The ratio of green and red signal reflects the relative mRNA expression levels. As shown in (b), yellow on the scanned slide represents equal expression in both samples. The hybridisations were done at least in triplicate for each group. No dye swap experiments were required as there was no direct comparison to the reference sample and all tumour cell lines being compared were always labelled with the same red Alexa Fluor®647 dye.

~~Chapter 2: Gene Expression Patterns in STS Progression~~

fibreoptic technology and a pulsed xenon flash lamp to provide readings which are accurate to within ± 1.5 ng/μl, up to a maximum concentration of 4000 ng/μl. 1 μl of sample is required per reading. An A260/A280 ratio of > 1.8 was used to determine and confirm the purity of the RNA sample.

~~2.2.6 cDNA Synthesis, Labelling and Hybridisation~~

The steps involved are schematically represented in Figure 2.1 and the detailed protocol is listed in Appendix I. Total RNA was reverse transcribed and labelled using SuperscriptII indirect labelling system (Invitrogen, Carlsbad, CA) and Alexa Fluor® 555 or Alexa Fluor® 647 as the fluorescent dyes (Invitrogen). These correspond to Cy3 and Cy5 (Amersham) respectively. No difference in signal intensity was observed between Cy dye and Alexa Fluor® labelling during initial optimisation of experimental protocols and Alexa fluors were subsequently used for all experiments. Alexa Fluor® 647 was used for labelling sarcoma cell line samples, while Alexa Fluor® 555 was used for labelling pooled MRC5.

The purified labelled cDNA was assessed on the NanoDrop, allowing for quality control. For labelled cDNA, the NanoDrop provided a measure of dye incorporation in pmol / μl, ensuring adequate and equivalent sample loading onto the microarray slide. 95 μl of the combined Alexa Fluor ® 555- and Alexa Fluor ® 647 labelled sample was ο loaded onto each 19K slide for hybridising overnight at 37 C in a humidified chamber.

~~2.2.7 Scanning~~

The slides were scanned using GenePix® 4000B laser scanner (Axon Instruments, Molecular Devices). This was done for each hybridised slide using standard protocols, adjusting the PMT (photomultiplier tube) gains in each channel (wavelength 635 nm and 532 nm) to avoid saturated spots and also to obtain count ratios of approximately 1.0 for red/green signals. The PMT gains used ranged from 650 - 700. The appropriate gene array list (gal) file, containing the gene identification for each spot according to its position on the array was loaded on the scanned image, to produce a “gpr” results file. Spots with abnormal morphology or those with inadequate data are automatically “flagged”. Additional manual inspection was carried out to exclude any other spots with abnormal morphology such as those described as donuts, tails and moons, as well as

~~Chapter 2: Gene Expression Patterns in STS Progression~~

oversaturated spots. Spots from areas of the array with defects were excluded from further analysis. The raw data in these “gpr” files were uploaded onto the GEO database along with experimental protocol details in accordance with MIAME compliance21.

~~2.2.8 Statistical Analysis of Microarray Data~~

~~2.2.8.1 GeneSpring~~

GeneSpring software (Silicon Genetics, Agilent Technologies, Palo Alto, CA) was used for further microarray analyses. Separate two way- (comparing the low and high grade tumour lines) and three way- (comparing all three tumour grades) analyses were carried out. Normalisation was used to remove process variation, reducing the 19 000 genes to some 17 500 genes. This “normalises” all signal(Cy3) / control(Cy5) ratios within one chip to a median of 1.0 (Quackenbush 2002; Yang, Dudoit et al. 2002). The data (signal intensity) were log2-transformed, and further filtering on confidence was carried out. A one way ANOVA (Kerr, Martin et al. 2000) was then applied to this quality controlled list to obtain a list of genes that were statistically differentially expressed across the groups. Two-dimensional hierarchical clustering was then applied to the list of statistically significant genes with Pearson correlation as well as k means clustering. k means randomly divides genes into a user defined number of clusters. The “centroid” for each cluster is identified. Reassignment of each gene to its closest centroid then occurs. Each graph in k means represents a cluster of genes with common expression patterns. The purpose of the various clustering methods used was to identify coregulated genes, genes in common pathways and find recurring patterns in the data.

21 The Gene Expression Omnibus (GEO) is a public repository of microarray data (Edgar, Domrachev et al. 2002; Barrett, Troup et al. 2007) and minimum information about a microarray experiment (MIAME) must also be provided (Brazma, Hingamp et al. 2001). 97

~~Chapter 2: Gene Expression Patterns in STS Progression~~

~~2.2.8.2 Comparison with Bioconductor R statistical package~~

Independent analysis of the raw data was carried out by Dr Rohan Williams (School of Biotechnology and Biomolecular Sciences, UNSW (Appendix I). Similar initial quality control measures such as exclusion of genes that had intensities less than the 95th percentile of the intensity values for the blank spots were used. Gene expression profiles for the top ranked genes were generated.

~~2.2.8.3 Gene Ontologies and Pathways~~

The above gene clusters were examined for their ontologies (Ashburner, Ball et al. 2000) and pathways using a combination of databases including National Institute of Health Database for Annotation, Visualisation and Integrated discovery (DAVID) (http://apps1.niaid.nih.gov/david/), Weizmann Institute of Health GeneCards database (http://bioinfo.weizmann.ac.il/cards/index.shtml), the Kyoto Encyclopaedia of Genes and Genomes or KEGG (http://www.genome.jp/kegg/pathway.html) and Gene Map Annotator and Pathway Provider or GenMAPP (http://www.genmapp.org/download.asp).

~~2.3 RESULTS~~

~~2.3.1 Cell Proliferation Rates~~

The results of the MTT assay performed on the four cell lines used in this part of the study were as follows: SW684: 45 h, HT1080: 36 h, GCT: 45 h and MRC5: 66 h. HT1080 had a faster proliferation rate than the metastatic cell line GCT. The patient from whom the high grade cell line HT1080 was derived did, at a later stage, progress to develop metastasis. Adjusting for their proliferation rates, the cell lines were all harvested for RNA extraction at 80 % confluence.

~~2.3.2 Assessment of RNA Quality and Labelled cDNA~~

All RNA samples were first assessed on the Agilent 2100 Bioanalyzer and NanoDrop ND-1000. An example of the results obtained on the Bioanalyzer is given in Figure 2.2. The samples were rechecked on the NanoDrop, which provided 260/280 ratios and RNA concentrations. There was good correlation between the two methods. RNA with 28S/18S ratios above 1.8, 260/280 ratios above 1.8 and concentrations greater than 1.3

~~Chapter 2: Gene Expression Patterns in STS Progression~~

~~μg / μl, was used for cDNA synthesis and labelling. No additional RNA amplification or concentration was required, as this may have introduced further variables into the experimental analysis.~~

The concentration and dye incorporation of labelled cDNA was assessed on the NanoDrop ND-1000 once the product had been purified. Two examples of results obtained are shown in Figure 2.3. These measurements were performed for every hybridisation as another method of quality control.

~~2.3.3 Scanning of 19K microarray slides~~

No normalisation was carried out within the GenePix software program as this was all done in GeneSpring. Slides with smears or high background were excluded completely from the analysis. As GeneSpring utilises the gene identities based on the gpr files, the different arrangements of the blocks, genes and control spots on the arrays from the two microarray facilities are accounted for.

~~Chapter 2: Gene Expression Patterns in STS Progression~~

Figure 2.2 Typical BioAnalyser (Agilent) result using the RNA 6000 Nano LabChip (a) Gel image generated. The first lane on this gel is the control ladder, the rest are samples. The smear in lane 11 corresponds to a degraded sample. Good quality RNA appears as flat baselines and tight peaks as on the corresponding electropherograms in (b), which are labelled with the sample cell line names and passage numbers. Genomic DNA contamination can result in broad peaks. 28S/18S ratios and RNA concentration in ng/μμμl are provided for each sample. RNA with concentrations > 1.25 μμμg/μμμl and 28S/18S ratios > 1.8 were used for cDNA synthesis. RNA from multiple harvests of MRC5, the normal fibroblast cell line (used as the reference green fluorescent labelled samples on all hybridisations) were first assessed on the Bioanalyzer, then pooled together. Aliquots from this pooled RNA were used on all the slides.

~~100~~

~~Chapter 2: Gene Expression Patterns in STS Progression~~

Figure 2.3 Assessment of Labelling Reaction The NanoDrop ND-1000 provided a measure of dye incorporation of labelled cDNA in pmol/μμμl, ensuring adequate and equivalent sample loading onto microarray slide. The spectrophotometer reading in (a) shows equivalent levels of dye incorporation in the Cy3 and Cy5 labelled samples, while (b) shows higher incorporation of the Cy5 labelled sample. The concentration of the combined cDNA is also given in ng/μμμl.

~~101~~

~~Chapter 2: Gene Expression Patterns in STS Progression~~

Figure 2.4 A scanned image of a hybridised 19K slide This slide was of green labelled MRC5 and red fluorescent labelled GCT, the metastatic cell line. Yellow spots indicate expression for that Oligonucleotide sequence in both the red and green labelled samples. The four subarrays in the top right hand corner have GAPDH printed diagonally as a quality control aid. The signal intensities for each spot in each channel is measured as a composite of the pixels that comprise it and numeric figures and ratios are calculated for this. Once the gal file is applied, the results are presented as a table format (gpr file) that is then imported into analysis programs such as GeneSpring.

~~102~~

~~Chapter 2: Gene Expression Patterns in STS Progression~~

~~2.3.4 Results of Analysis comparing Low and High grade cell lines~~

~~Tables of gene lists are placed in Appendix II, due to their length. Figures are incorporated within the body of the text.~~

~~2.3.4.1 GeneSpring~~

There were 414 genes found to be statistically differentially expressed using one way ANOVA. Some of the genes that were overexpressed on hierarchical clustering in the high grade cell line HT1080 relative to the low grade SW684 are given in Table II.1 (Appendix II).

2.3.4.1.1 TGF! Family and Associated Factors Genes from the transforming growth factor beta (TGF!) family (TGFB1, ACVR2B and LTBP3), as well as some cell cycle regulators (CDKN2D, CDNK2A, CCNB2, CCNB1IP1) were differentially expressed. Cyclin B2, which binds to transforming growth factor receptor beta receptor II, is thought to play a role in TGF! mediated cell cycle control. Members of the signal transduction pathway involving the interleukin 1 receptor (IL1R), such as interleukin 1 beta (IL1B), interleukin 1 receptor accessory protein (IL1RAP) and mitogen-activated protein kinase kinase kinase 14 (MAP3K14, alias NIK) were also overexpressed. It is of note that TGF! 1 is also able to signal through the IL1R and MAPK pathways.

2.3.4.1.2 Hypoxia-Inducible Factor Pathway Genes involved in the hypoxia-inducible factor pathway were also upregulated in the high grade cell line but not the low grade cell line.. These upregulated genes included lactate dehydrogenase a (LDHA), nitric oxide synthase 3 (NOS3), eosinophil peroxidase (erythropoietin, EPX or EPO) and aspartate beta-hydroxylase (ASPH). N-myc downstream regulating gene 1 (NDRG1), a member of the alpha/beta hydrolase superfamily known to be modulated by hypoxia (Park, Adams et al. 2000; Cangul 2004), was also overexpressed in the high grade sarcoma cell line HT1080.

2.3.4.1.3 Cell Adhesion and Migration Genes involved in cell adhesion and breakdown of extracellular matrix (ECM), such as matrix metalloproteinases (MMPs), integrins and laminins (MMP14, ITGA2, ITGA6 and LAMB3) were also shown to be differentially expressed.

~~103~~

~~Chapter 2: Gene Expression Patterns in STS Progression~~

2.3.4.1.4 Other Significant Upregulated Genes A number of tyrosine and serine/threonine kinases, G proteins and factors with EGF- motifs such as TIE1 and DNER were also upregulated in the high grade cell line.

The corresponding gene ontologies for this set of genes is shown in Figure 2.5. This was carried out for the categories of biologic process and molecular function, specifying a minimum of 3 genes per function or process and an intermediate level of stringency for both coverage and specificity. This allows the general pattern of involvement of the functional groups to be appreciated.

~~2.3.4.2 R analysis~~

317 genes were found to be statistically differentially expressed between the high and low grade cell lines. The top 170 genes with B >1 are presented in Table II.2, in Appendix II. The difference in expression, ΔΔΔM (HT1080-SW684) is shown in the last column of the table. A value of 1 represents a two fold increase in expression in the high grade cell line HT1080 compared to the low grade SW684, while a value of − 1 refers to a two fold overexpression in the low grade cell line relative to the high grade. A volcano plot (Log Odds plot) of the B value against the difference in M (log2 red / green signal for each gene between the sarcoma cell lines) is shown in Figure 2.6.

Again, there are genes overexpressed in the high grade sarcoma cell line involving growth factor signal transduction pathways (TGFBI, IGFBP7, PRKCD, among others), cell cycle regulation (such as CCND1), cell adhesion and migration (such as the integrins ITGA2 and ITGA6, and LAMB3). IGFBP7 has been shown in other studies to be involved in breast, bladder and gastric cancer as well as sarcomatoid malignant mesothelioma (Kim, Bang et al. 2003; Bieche, Lerebours et al. 2004; Xiaojuan Sun 2005; Osman, Bajorin et al. 2006). There was downregulation of ABCC2, a member of the ATP binding cassette family thought to be involved in drug resistance, in the high grade sarcoma cell line HT1080. There was also differential expression of genes in the Wnt-FZD pathway, such as the alpha-like catenin (CTNNAL1), wingless-type MMTV integration site member 5B (WNT5B) and disheveled associated activator of morphogenesis 1 (DAAM1). WNT5B is

~~104~~

~~Chapter 2: Gene Expression Patterns in STS Progression~~

relatively underexpressed in the high grade sarcoma line in contrast to the other abovementioned genes which were overexpressed. Cancer testis antigen 1 (CTAG1) or LAGE2A is also overexpressed in the high grade sarcoma line HT1080. This has been previously noted by Maio et al (Maio, Coral et al. 2003), when investigating the expression of CTAG1 in medullary thyroid cancers. CTAG1 expression has also been reported in oesophageal cancers and melanoma.

~~Gene Ontology Genes Overexpressed in High Grade Sarcoma Cell Line versus Low Grade~~

~~3 3 3 3 3 33 27 4 4 4 4 4 27 5 4 5 5 5 6 25 6 6~~

~~7 21 7 9 9 14 9 14 9 13 9 9 13 10 10 13~~

cell growth and/or maintenance protein metabolism signal transduction nucleobase, nucleoside, nucleotide and nucleic acid metabolism DNA binding biosynthesis purine nucleotide binding response to external stimulus transferase activity, transferring phosphorus-containing groups RNA binding phosphorus metabolism catabolism cell death cell-cell signaling organogenesis protein kinase activity response to stress cytoskeletal protein binding energy pathways amine metabolism hydrolase activity, acting on ester bonds organic acid metabolism amino acid and derivative metabolism carbohydrate metabolism translation factor activity, nucleic acid binding transmembrane receptor activity GTPase activator activity cell adhesion growth factor activity immune response regulation of cell proliferation small GTPase regulatory/interacting protein activity calmodulin binding cell motility cytokine activity electron transport lipid metabolism regulation of body fluids translation initiation factor activity

~~105~~

~~Chapter 2: Gene Expression Patterns in STS Progression~~

Figure 2.5 Gene Ontology for genes overexpressed in the High grade cell line The databases classify genes according to the biologic process, molecular function and cellular component involved. The gene ontology database used for the classification below was accessed through the National Institute of Allergy and Infectious Diseases (NIAID) web based Database for Annotation, Visualisation and Integrated Discovery (DAVID): http://apps1.niaid.nih.gov/david/ There is significant representation in the signal transduction, while the greatest number of genes is seen in the categories of cell growth and protein metabolism.

~~'$$#0#,!# ', %0-3. +# , %3SRZR 41 84#0 %# .#0V%#,# 8~~

~~84#0 %# .#0V%#,# 8~~

~~Figure 2.6 Volcano plots generated with Bioconductor R package~~

~~106~~

~~Chapter 2: Gene Expression Patterns in STS Progression~~

The above compares the high and low grade cell lines. In graphs (a) and (b), the y-axis shows the gene ranking system expressed as the B statistic. A similar ranking system is used for the 3 way comparison, represented as the Moderated F statistic on the y-axis in graph (c). The x-axis in graph (a) represents the difference in gene expression between the high and low grade cell lines and in graphs (b) and (c), the average intensity of expression for each gene across all samples. Each dot on the graphs represents a gene. The position of selected genes on the plots is shown.

~~Gene Ontology Genes Differentially Expressed between High and Low Grade Sarcoma Cell Lines~~

~~3 3 33 3 3 3 3 34 4 4 4 4 4 4 25 5 5 5 5 18~~

~~5 6 6 13 6 7 13 8 8 12 8 10 10 10 10~~

cell growth and/or maintenance signal transduction protein metabolism nucleobase, nucleoside, nucleotide and nucleic acid metabolism response to external stimulus cell adhesion biosynthesis catabolism organogenesis purine nucleotide binding DNA binding peptidase activity transferase activity, transferring phosphorus-containing groups lipid metabolism calcium ion binding hydrolase activity, acting on acid anhydrides immune response cell death cell-cell signaling hydrolase activity, acting on ester bonds phosphorus metabolism response to stress cytoskeletal protein binding growth factor binding neurophysiological process organic acid metabolism protein kinase activity regulation of cell growth RNA binding carbohydrate metabolism cytokine activity growth factor activity magnesium ion binding organismal movement regulation of cell proliferation transition metal ion binding Figure 2.7 Gene Ontology for differentially expressed genes on Bioconductor R The corresponding list of 170 genes is given in Table II.2 in Appendix II. A number of genes remained unclassified due to the criteria specified, such as exclusion of categories for which there were fewer than 3 genes in the list and intermediate levels of specificity and coverage. Signal transducers and growth factors are differentially expressed, as previously noted. Of interest also is a number of genes involved organogenesis. Genes usually involved in development can be reactivated in malignancy. This is examined in greater detail in the discussion section of the chapter.

~~107~~

~~Chapter 2: Gene Expression Patterns in STS Progression~~

~~2.3.5 Results of Analysis comparing all three groups~~

~~2.3.5.1 GeneSpring~~

Some of the differentially expressed genes is shown in Table II.3. Hierarchical clustering, based on centroid linkage, is shown in Figure 2.8, providing a representation of the differential expression patterns. Portions of the dendrogram can be examined in greater detail, creating “subtrees” or clusters of genes, as shown in Figure 2.8 which demonstrates increasing levels of expression for the genes from the low grade to the metastatic cell line. Hence the genes in this sub-cluster may be involved in tumour progression and metastasis. The corresponding gene ontologies (GO) for this group are represented in Table II.5 and Figure 2.9. The genes in this subtree are listed fully in Table II.4 in Appendix II. The converse is represented in Figure 2.10, with high gene expression levels in the low grade cell line and lower expression in the high grade and metastatic lines. This group of genes may be markers of favourable prognosis. K means clustering was also carried out, specifying 5 clusters as the user defined criteria (Figure 2.11). There was good correlation between the genes in Set 1 in k means clustering and the subtree detailed in Figure 2.8.. All 113 genes in the subtree showing increased expression levels in the high grade and metastatic groups were present in Set 1 (Table II.6), as defined by the k means clusters.

2.3.5.1.1 EGF Family Of note, genes found to be over-expressed in the high grade and metastatic cell lines included those in the epidermal growth factor (EGF) family such as EGFR and its ligand epiregulin (EREG) (Tables II.3, II.7), and other molecules with EGF-motifs such as DNER (Tables II.4, II.6) and ELTD1 (Table II.6). These are in addition to another EGF-motif containing molecule EFEMP1 being found to be over-expressed in the high grade sarcoma cell line (Table II.2). Downstream signal transducers of EGFR, including those in the MAP kinase cascade, MAP3K7 and MAPK13 (Table II.4) were similarly over-expressed in the high grade and metastatic lines.

~~2.3.5.1.2 TGF! Family As in the previous 2-group analysis, certain members of the TGF! family were also upregulated in the high grade and metastatic sarcoma cell lines. These included TGF! -~~

~~108~~

~~Chapter 2: Gene Expression Patterns in STS Progression~~

~~induced factor 2 (TGIF2) (Table II.6), TGF! -induced (TGFBI) (Table II.3) and the TGF! signal transducer MADH5 / SMAD5 (Table II.6).~~

2.3.5.1.3 Other Genes and Pathways Genes involved in cell adhesion and breakdown of ECM such as MMP1 and ICAM1 (Table II.3) were upregulated in the higher grade cell lines, as were genes involved in the developmental Notch pathway (NOTCH4, DNER) (Tables II.4, II.6). NDRG1 (Table II.3) was upregulated, as was VEGFC (Table II.7).

~~109~~

~~Chapter 2: Gene Expression Patterns in STS Progression~~

Figure 2.8 Hierarchical clustering of gene expression data The branches of the dendrogram at the top are colour coded for each tumour grade (red: low grade; yellow: metastatic; turquoise: high grade). Each column represents a single hybridisation while each row is a gene. Each bar in the rows is the colour coded representation of expression level as a ratio of tumour cell signal intensity to reference MRC5 cDNA intensity. Red signifies relative overexpression and green, underexpression. The replicates group together, as shown by the relatedness of the top dendrogram and differential gene expression patterns are evident across the 3 groups. The horizontal clustering reveals groups of genes with similar expression patterns, potentially identifying coregulated genes. The left panel displays the entire heatmap constructed with the 883 filtered and statistically differentially expressed genes. The right panel is an enlargement of a portion of the tree displaying genes that are overexpressed in the high grade and metastatic cell lines. These genes may be involved in progression of disease.

~~110~~

~~Chapter 2: Gene Expression Patterns in STS Progression~~

~~Gene Ontologies Increased gene expression with Tumour Grade~~

adenyl nucleotide binding transcription cell proliferation macromolecule biosynthesis transport protein biosynthesis cell surface receptor linked signal transduction intracellular signaling cascade protein modification RNA metabolism phosphate metabolism phosphotransferase activity, alcohol group as acceptor response to biotic stimulus protein serine/threonine kinase activity guanyl nucleotide binding nucleotidyltransferase activity protein targeting response to pest/pathogen/parasite DNA metabolism G-protein coupled receptor activity GTPase activity amino acid metabolism amino sugar metabolism carboxylic acid metabolism detection of external stimulus guanyl-nucleotide exchange factor activity humoral immune response hydrolase activity, acting on acid anhydrides, in phosphorus-containing anhydrides innate immune response macromolecule catabolism negative regulation of cell proliferation negative regulation of metabolism nucleotide metabolism oxidoreductase activity, acting on the CH-OH group of donors, NAD or NADP as acceptor programmed cell death protein catabolism protein folding protein-tyrosine kinase activity regulation of biosynthesis response to DNA damage stimulus response to wounding sensory perception transcription factor activity transferase activity, transferring hexosyl groups zinc ion binding

Figure 2.9 Gene ontologies of genes increasing in expression with increasing tumour grade The ontologies were obtained from the web-based site of the National Institute of Allergy and Infectious Diseases (NIAID) Database for Annotation, Visualisation and Integrated Discovery (DAVID) (http://apps1.niaid.nih.gov/david/). The unclassified genes are excluded from this representation. The more stringent the criteria applied to the gene ontology, the more specific the categorisation but the greater the number of genes that remain unclassified. Significant proportions of classified genes are involved in cell surface and intracellular signal transduction, serine/threonine kinase activity, ATP binding, transcription and cell proliferation. (See also Tables II.4 and II.5 (Appendix II) and Figure 2.8).

~~111~~

~~Chapter 2: Gene Expression Patterns in STS Progression~~

~~Low High Metastatic~~

Figure 2.10 Hierarchical clustering of gene expression data The branches of the dendrogram at the top are colour coded for each tumour grade (red: low grade; yellow: metastatic; turquoise: high grade). As before, each column represents a single hybridisation or sample, while each row is a gene. Red signifies relative over-expression and green, under-expression. Colour saturation is proportional to the magnitude of the difference from the mean. The left panel displays the entire heatmap constructed with the 883 filtered and statistically differentially expressed genes. The middle panel is an enlargement of a portion of the tree displaying genes that are decreasing in expression from the low grade to the metastatic cell lines. The box on the right lists some of those genes by their GenBank accession numbers and gene symbols. These genes may therefore be associated with a better clinical prognosis.

~~112~~

~~Chapter 2: Gene Expression Patterns in STS Progression~~

Figure 2.11 Results of k means clustering 5 groups specified. Each graph in k means represents a cluster of genes with common expression patterns. Tumour grades are on the x-axis & normalised log ratios (gene expression levels) on the y- axis. Sets 1, for example, comprises a cluster of genes that show progressively increased expression with tumour grade while Set 5 consists of a group of genes overexpressed in the low grade cell line. Set 1 may therefore identify genes involved in tumour progression and metastasis and Set 5 identify genes that protect against it. Set 3 contains a cluster overexpressed in the metastatic cell line. The genes in Sets 1 and 3 are listed in Table II.6 and Table II.7 in Appendix II. The gene ontology for Set 1 is shown in Appendix II.

~~113~~

~~Chapter 2: Gene Expression Patterns in STS Progression~~

~~2.3.5.2 R analysis~~

A comparison across the three cell lines was carried out as described. A moderated F statistic was used to define the differentially expressed genes across the groups. The top 60 genes from this analysis are shown in Table II.8 (Appendix II). It included genes involved in cell motility, invasion and metastasis such as Integrins alpha 2 and 6 (ITGA2 and ITGA6), the extracellular matrix component Laminin B3 (LAMB3), collagen type VI alpha 3 (COL6A3) and matrix metalloproteinase 3 (MMP3). These are considered further in the discussion. Genes involved in the apoptosis pathway were differentially regulated. These included BID, NFKBIA as well as IL1A and IL1B.

As with the analysis done on GeneSpring, N-myc downstream regulating gene 1 (NDRG1) was overexpressed in the higher grade cell lines compared to the low grade. NDRG1 in normal tissue is thought to be involved in growth arrest and differentiation, and in peripheral nerves, is expressed in Schwann cells (Kalaydjieva, Gresham et al. 2000). Its expression in malignancy is considered further in the discussion of candidate genes in Section 2.4.4.3.7.

~~2.4 DISCUSSION~~

The earliest gene expression studies began to be reported in the late nineties. Now, nearly a decade on, many authors have reviewed the successes and limitations of technique (King and Sinha 2001; Miller, Long et al. 2002; van de Rijn and Rubin 2002; Bertucci, Viens et al. 2003; Nilbert, Meza-Zepeda et al. 2004). Numerous sources of variability exist at each step of a microarray study, from RNA extraction to data analysis and there is a need for standardisation of protocols if meaningful comparisons of microarray datasets are to be made from different groups of researchers. In terms of the present study, the options and rationale employed are discussed in Appendix I. In the following sections, the candidate genes and pathways involved in tumour progression in sarcoma are considered, together with the use of cell lines for this process of gene discovery.

~~2.4.1 The Use of Cell Lines for Gene Profiling~~

~~There are limitations of using cell lines as a model for any gene expression or functional study, the most common criticism being that the cells will inevitably have changed with~~

~~114~~

~~Chapter 2: Gene Expression Patterns in STS Progression~~

subculturing and may no longer reflect the gene expression patterns of the tissue of origin. However the following difficulties need to be acknowledged in this context: • It would be virtually impossible to harvest tumour from the same patient of varying histologic grade. Low grade tumours do not necessarily progress to high grade tumours. It would of course also be unethical to leave a clinically evident tumour untreated to follow its natural history. While a patient may present initially with a high grade tumour and then progress to develop metastases despite treatment, the chances of successfully establishing cell lines from both these tumours would be very low, given the inherent difficulty of this process and the time required to establish cell lines. • In view of the limitations of interpretation in using cell lines, it would have been preferable to use soft tissue sarcoma samples of varying histologic grade for the gene expression arrays. However the relative rarity of soft tissue sarcomas meant that the collection of fresh tumour specimens and establishing a tumour bank in a single institution was a slow process and the time constraints of a PhD project would not have allowed for the accumulation of sufficient specimens for the study. Paraffin embedded archival specimens could have been used, but the quality of RNA obtained from such specimens is inevitably poor and would not have been suitable for gene expression arrays. Techniques such as laser capture microdissection (LCM) have been employed in various studies, with the aim of obtaining only tumour cells with the exclusion of adjacent stroma (Okabe, Satoh et al. 2001; Ma, Salunga et al. 2003; Prasad, Biankin et al. 2005) This inevitably necessitates amplification of the small amount of RNA extracted. Whether certain transcripts are preferentially amplified during this process is not known but remains a potential source of bias. It is therefore preferable not to use amplified RNA in microarray experiments.

Cell lines, as used in the present study, are an attractive option, due to the relative ease with which good quality and abundant RNA can be extracted from them. Previous studies have, in addition, demonstrated the validity of using cell lines for the discovery of prognostic markers (Han, Bearss et al. 2002; Sakakura, Hagiwara et al. 2002; Sanchez-Carbayo, Socci et al. 2002). A cogent argument for the use of cell lines is

~~115~~

~~Chapter 2: Gene Expression Patterns in STS Progression~~

presented by Ross et al (Ross, Scherf et al. 2000) , where the expression patterns of 60 cell lines as well as fresh frozen normal and tumour tissue were examined and hierarchical clustering performed:

9-*=57.2&7>=.)*39.+.&'1*=+&(947=&((4:39.3,=+47=;&7.&9.43=.3=,*3*=*=57*88.43= 5&99*738=&243,=9-*8*=0*=(*11=1.3*8=<&8=9-*=.)*39.9>=4+=9-*=9.88:*=+742=<-.(-= *&(-= (*11= 1.3*= <&8= 489*38.'1>= )*7.;*)c_3*.9-*7= 5->8.414,.(&1= 347= *=5*7.2*39&1= &)&59&9.43= +47= ,74<9-= .3= (:19:7*= <&8= 8:++.(.*39= 94= 4;*7<7.9*= 9-*=,*3*=*=57*88.43=574,7&28=*89&'1.8-*)=):7.3,=).++*7*39.&9.43=.3=;.;4_= Analysis of tumour grade, rather than histologic classification per se holds merit. An earlier gene expression study by Ramaswamy et al (Ramaswamy, Tamayo et al. 2001), attempting to cluster tumours by their cell of origin, found in fact, that the high grade tumours would not fall into the expected clusters:

9-*= 54471>= ).++*7*39.&9*)= 9:2478= &3&1>?*)= .3= 9-.8= 89:)>= (4:1)= 349= '*= (1&88.+.*)=&((47).3,=94=9-*.7=9.88:*8=4+=47.,.3____4:7=)&9&=.3).(&9*=9-&9=54471>= ).++*7*39.&9*)=9:2478=-&;*=&=;*7>=).++*7*39=,*3*=*=57*88.43=574,7&2_=c_=9-.8= +.3).3,=7&.8*8=9-*= 5488.'.1.9.*8=9-&9= 54471>=).++*7*39.&9*)=9:2478=&7.8*= +742= ).89.3(9= (*11:1&7= 57*(:78478`= -&;*= ).++*7*39= 241*(:1&7= 2*(-&3.828= 4+= 97&38+472&9.43`= 47= -&;*= :3.6:*= 3&9:7&1= -.8947.*8= .3= 842*= 49-*7= 7*85*(9c_= .;*3=9-*=(1.3.(&11>=&,,7*88.;*=3&9:7*=4+=54471>=).++*7*39.&9*)=(&3(*78`=842*= 2&70*78= 4+= 54471>= ).++*7*39.&9*)= 9:2478= 2.,-9= 574;*= ,*3*7&11>= :8*+:1= +47= 57*).(9.3,=5447*7=(1.3.(&1=4:9(42*_= There are molecular changes that occur during the metastatic cascade that are common to all cancers, such as loss of contact inhibition, destruction of the extracellular matrix and invasion. It is reasonable to assume that these abilities are characterised by particular gene expression patterns that hold the key to prognostics and therapeutic targeting. This supports the rationale, for employing the cell lines of increasing grade or metastatic potential as a model for generating hypotheses and identifying markers that warrant further attention.

Understanding the limitations of using three cell lines, validation of the gene expression data was thus of paramount importance. The gene expression study was used as a means of generating hypotheses, rather than drawing firm conclusions. These hypotheses were then tested further by first performing quantitative RT-PCR in real

~~116~~

~~Chapter 2: Gene Expression Patterns in STS Progression~~

~~time, and subsequently examining protein expression of selected candidates on paraffin embedded soft tissue sarcoma samples on a tissue array.~~

~~2.4.2 Candidate Genes and Associated Pathways~~

~~2.4.2.1 EGFR~~

Epidermal growth factor receptor (HER1, EGFR or ErbB-1) was found to be most highly expressed in the metastatic sarcoma cell line, as was its ligand epiregulin (EREG). Other genes with EGF-motifs (EFEMP1, TIE1, ELTD1 and DNER) were also upregulated in the higher grade cell lines. EGFR is a transmembrane tyrosine kinase receptor. Binding of any of a number of ligands, including EGF, TGF-α, EGF-like molecules, neuregulins and epiregulin activates the receptor inducing the formation of receptor homodimers or heterodimers with other members, HER-2/neu (ErbB-2), HER- 3 (ErbB-3) and HER-4 (ErbB-4), triggering tyrosine kinase enzymatic activity. Downstream signal transduction classically involves the Sos-Ras-Raf and mitogen- activated protein kinase pathway, which in turn modulates cell proliferation, survival, adhesion, migration and differentiation (Section 2.4.5.4) (Ciardiello and Tortora 2001; Yarden 2001). EGFR has been implicated in many cancers, including nonsmall cell lung cancer, breast cancer, pancreatic cancer and glioblastoma (Hirsch, Varella-Garcia et al. 2003; Mukohara, Kudoh et al. 2003; Thomas, Chouinard et al. 2003; Baselga, Albanell et al. 2005; El-Rayes, Ali et al. 2006). This has rapidly led to the development of EGFR and other broad spectrum tyrosine kinase inhibitors that have already been applied to clinical trials in colon, breast and non small cell lung cancer (NSCLC), among others (Ciardiello and Tortora 2001; Ciardiello, De Vita et al. 2004; Baselga, Albanell et al. 2005). EGFR expression in soft tissue sarcoma has been reported in gene expression studies, mainly in relation to synovial sarcoma (Allander, Illei et al. 2002; Nielsen, West et al. 2002; Baird, Davis et al. 2005). Another gene expression study did not, however, report an overexpression of EGFR in synovial or other sarcomas (Nagayama, Katagiri et al. 2002).

~~117~~

~~Chapter 2: Gene Expression Patterns in STS Progression~~

The EGF-like molecule, epiregulin binds to EGFR with less affinity than EGF. It is rarely expressed in normal adult tissue but has been reported in many epithelial cancer cell lines (Toyoda, Komurasaki et al. 1997; Baba, Shirasawa et al. 2000; Torring, Jorgensen et al. 2000; Sorensen, Torring et al. 2004). An increased expression of epiregulin was noted in fibroblasts immortalised by hTERT, the catalytic subunit of the enzyme telomerase on gene expression profiling (Lindvall, Hou et al. 2003). It has also been proposed that epiregulin secreted by vascular smooth muscle contributes to vascular smooth muscle proliferation (Taylor, Cheng et al. 1999). This study showed an increasing level of expression of EGFR with histologic grade on gene expression profiling. The abovementioned studies have concentrated on synovial sarcomas, which by definition were high grade. Epiregulin (EREG) was also overexpressed in the metastatic cell line, GCT, together with TERT, indicating the coexpression and possible co-regulation of both telomerase and epiregulin in this instance. The expression of epiregulin in soft tissue sarcomas is yet to be determined. EGFR presents as an attractive prognostic and therapeutic target as drugs targeting the receptor, such as Z1839 (Gefitinib, Iressa), Erlotinib (Tarceva/ OSI-774) and the broader spectrum anti-VEGFA and EGFR inhibitor ZD6474 are already being trialled for other cancers. Whether EGFR is a prognostic marker for synovial sarcomas alone, or soft tissue sarcomas in general, remains to be confirmed. The further validation of these results is described in Chapter 3.

~~2.4.2.2 The TGF! Family and Relevant Pathways~~

A number of genes in this pathway (ACVR2B, TGFB1, TGIF2, LTBP3, SMAD5/MADH5, TAK1/MAP3K7) were shown to be over-expressed in the high grade and/or metastatic sarcoma cell lines compared to the low grade (Figure 2.12). Transforming growth factor beta (TGF!) was originally named for its ability to transform and cause proliferation of mesenchymal cells. In epithelial cells, it was thought to have growth inhibitory effects but epithelial cancers, such as colon, breast and prostate cancer, are known to become insensitive to the growth inhibitory effects of TGF!. In some instances, this may be through downstream mutations, as in SMAD4 in pancreatic cancer (Hahn, Schutte et al. 1996). The role of Smad-independent pathways

~~118~~

~~Chapter 2: Gene Expression Patterns in STS Progression~~

in TGF! signalling in cancer has been explored (Yu, Hebert et al. 2002; Derynck and Zhang 2003; Roberts and Wakefield 2003; Elliott and Blobe 2005). These include direct effects on integrins, as well as the Erk, JNK and p38 MAPK kinase pathways. It has been proposed that in epithelial tumours, the role of TGF! in tumour progression is effected through an epithelial to mesenchymal transition or EMT22 (Thiery 2002; Elliott and Blobe 2005) (Figure II.4, Appendix II). It is possible that these Smad - independent mechanisms are operating in sarcomas. This may explain the overexpression of TGFB1, TAK1(MAP3K7) and p38delta (MAPK13) but not the majority of the intracellular Smads in the high grade sarcoma cell line on the microarrays used in the present study. Interestingly, a recent gene expression study noted overexpression of TGFB1, SMAD6 and BMP7 in synovial sarcomas (Francis, Namlos et al. 2007). Activins (also known as inhibins) are dimeric growth and differentiation factors which belong to the TGF! superfamily and share extensive sequence homology with the TGF! sequence. Inhibin A, comprised of two alpha subunits, is normally produced by ovarian granulosa cells. Inhba-null mice demonstrate disruption of palate and tooth development, while homozygous Inhbb-null mice have fertility and eye defects (Brown, Houston-Hawkins et al. 2000). Activin signalling occurs through a heteromeric complex of transmembrane receptor serine kinases which include at least two type I (I and IB) and two type II (II and IIB) receptors. ACVR2B is thought to play a critical role in patterning both anteroposterior and left-right axes in vertebrate animals (Oh and Li 1997). It is thought that the ligands that signal through Activin type 2 receptors during embryonic development are not activins (Matzuk, Kumar et al. 1995). Activins and activin receptors have been studied in the context of cancer. Inhibin is produced by granulosa-cell tumors and is a useful early marker for primary, recurrent, and residual granulosa-cell tumors (Lappohn, Burger et al. 1989). Interestingly, as with TGF!, the

~~22 Epithelial-mesenchymal transition and the converse phenotypic switch was introduced in Chapter One: Molecular Biology of Soft tissue Sarcomas 119~~

~~Chapter 2: Gene Expression Patterns in STS Progression~~

dual role of inhibins as both tumour suppressor in early stage cancer and metastasis promoter in advanced cancer has been proposed in the case of prostate cancer (Ball, Mellor et al. 2004; Risbridger, Ball et al. 2004). ACVR1B has recently been found to be differentially expressed between primary and metastatic leiomyosarcomas (LMS) (Lee, John et al. 2004), and ACVR2A, over-expressed in gastrointestinal stromal tumours (GISTs) (Segal, Pavlidis et al. 2003). In this study, there was overexpression of ACVR2B in the high grade sarcoma cell line compared to the low grade. TGFB1 and ACVR2B are examined further in Chapter 3, where transcript expression of these genes using quantitative qRT-PCR (qRT-PCR) is described.

~~2.4.2.3 N-myc downstream regulating gene 1~~

NDRG1, also known in the literature as DRG1, PROXY1 (Protein Regulated by OXYgen-1) and cap43 is expressed in a variety of normal epithelial cells such as the prostate, colon, small intestine, liver and pancreas. It is postulated to be involved in protection of tissue from ischaemic damage, with increased expression and nuclear localisation noted in hypoxia (Lachat, Shaw et al. 2002).

In epithelial tumours, NDRG1 overexpressed in oral squamous cell carcinoma (Chang, Wang et al. 2005), non-small cell lung cancers on cDNA microarrays (Kikuchi, Daigo et al. 2003) and colorectal cancer lymph node metastases (Wang, Wang et al. 2004). Conversely, an inverse correlation between NDRG1 expression and Gleason grade and prognosis was found in prostate cancer (Bandyopadhyay, Wang et al. 2006), with a similar inverse correlation noted between the biomarker and disease free survival from breast cancer (Bandyopadhyay, Pai et al. 2004). NDRG1 has also been reported to be downregulated in N-myc expressing neuroblastomas (Li and Kretzner 2003). NDRG1 expression has not been investigated in adult soft tissue sarcomas. In this thesis, NDRG1 expression was found to be increased in the high grade cell line and its expression in sarcoma cell lines used in this study was further validated using qRT-PCR in Chapter 3.

~~2.4.2.4 Cell cycle regulators~~

~~Many of the ligands, receptors and signalling cascades have downstream effects on cell cycle survival and apoptosis. Cell cycle regulators were also found in this study to be~~

~~120~~

~~Chapter 2: Gene Expression Patterns in STS Progression~~

Figure 2.12 Differentially expressed genes in the TGF! signalling pathway. The differentially expressed genes are circled in red. TGFB1 and Activin receptor type 2B (ACVR2B) are overexpressed in the high grade sarcoma cell line compared to the low grade. Activins share sequence homology with TGF!, and the receptors, like the TFB receptors, are serine-threonine protein kinases. Classical TGF! signalling involves binding of TGF! in the extracellular space to either TGF! -receptor II or III, complexing with receptor I and signal transduction via the SMADs, particularly SMADs 2, 3 and 4. SMAD5, involved in BMP signal transduction, is overexpressed in the high grade cell line. TAK1 or MAP3K7, is also overexpressed, indicating upregulation in the mitogen activated protein kinase (MAPK) signalling pathway, a downstream target of TGF!. Recent literature has reported SMAD-independent pathways being involved in the tumour promoting effects of TGF!, including direct signalling via the MAPK pathway. This figure was adapted from the Kyoto Encyclopaedia of Genes and Genomes or KEGG database (http://www.genome.jp/kegg/pathway.html) dysregulated, with a complex picture emerging of the interplay of cell proliferation promoters and inhibitors and pro-and anti-apoptotic factors. Dysregulation of cyclins B2 and D1 (CCNB2 and CCND1), as well as cyclin dependant kinase inhibitors such as 2D and 1A (CDKN2D/p19 and CDKN1A/p21/Cip1/Waf1) was identified. Other overexpressed genes involved in cell cycle regulation in the high grade cell lines

~~121~~

~~Chapter 2: Gene Expression Patterns in STS Progression~~

included the mitosis inducer, CDC25C, the gene encoding BCL2-interacting protein that protects against cell death, BNIP1, and the BH3-interacting domain death agonist known as BID, the protein product of which interacts with either BAX or BCL2. The expression of CCND1 in sarcoma cell lines at mRNA transcript level is examined further in Chapter 3 with qRT-PCR.

~~2.4.2.5 The Mitogen Activated Protein Kinase (MAPK) Pathway~~

The MAPK signalling cascade is the integration point for a plethora of biochemical signals from growth factors, cytokines to physiologic stressors. It is involved in numerous processes such as cellular proliferation, cell survival, transcriptional regulation and development. The three main pathways of this cascade are shown in Figure 2.14. TGF-beta-activated kinase (TAK1 or MAP3K7), which has a putative N-terminal protein kinase domain, regulates transcription by transforming growth factor-beta (Dempsey, Sakurai et al. 2000). The protein encoded by MAP4K4 (also known as HGK) is a member of the serine/threonine protein kinase family and has been shown to specifically activate MAPK8/JNK and is thought to function through the MAP3K7-MAP2K4- MAP2K7 kinase cascade, and mediate the TNF-alpha signalling pathway. MINK 1 belongs to the germinal center kinase (GCK) family and is structurally similar to the kinases that are related to NIK. It activates the cJun N-terminal kinase (JNK) and the p38 pathways and may be involved in central nervous system development. p38 has been shown to have opposing effects on endothelial cells (McMullen, Bryant et al. 2005). It causes cell cycle arrest, limiting proliferation, but promotes cell migration by effecting secondary changes in actin architecture, thus increasing formation of lamellopodia. These MAP kinases were overexpressed (circled in Figure 2.14) in the high grade and metastatic sarcoma cell lines, together with the upstream ligands, TGFB1, IL1B and EGFR. EGFR activates the classical Sos-Ras-Raf MAPK cascade, as explained in Figure 2.14. The overexpression of one of these MAP kinases in high grade sarcoma, MAP4K4, is validated in Chapter 3 by qRT-PCR.

~~122~~

~~Chapter 2: Gene Expression Patterns in STS Progression~~

~~2.4.2.6 Cell Motility, Invasion and Angiogenesis~~

Laminins, integrins and matrix metalloproteinases (MMPs) are expressed during the process of invasion, metastasis and angiogenesis (Hood and Cheresh 2002)23. The tumour cell with the “metastatic phenotype” penetrates the extracellular matrix (ECM).

Integrins are the heterodimeric cell surface receptors that bind to ECM components such as laminins and fibronectin. This triggers a signalling cascade (Figure II.3, Appendix II) that alters the actin cytoskeleton. Adhesion promotes endothelial cell proliferation and motility during new blood vessel growth (Kim, Bell et al. 2000). Studies of melanomas, gliomas and endothelial cells have suggested that integrins may be part of the complex that activates MMPs, facilitating ECM degradation (Hood and Cheresh 2002). Integrins also transduce intracellular signals that promote cell migration and suppress apoptosis. They activate signalling pathways by activating focal adhesion kinases, or FAK (Jin and Varner 2004).

Gene expression profiling of the sarcoma cell lines in this study, showed increasing expression levels of Integrin alpha 2 and 6 (ITGA2 and ITGA6) from the low grade to the high cell line, as well as increasing expression of Laminin B3 (LAMB3), which binds to these integrins. This concurs with other studies which have linked the expression of integrins with invasion, metastasis and cell survival in breast and oral cancer (Shaw, Chao et al. 1996; Garzino-Demo, Carrozzo et al. 1998). This has implications for targeted therapeutics, particularly in light of the fact that antagonists of integrins α5β1 and αvβ3 and peptide inhibitors of integrins αvβ3 /αvβ5 are currently undergoing evaluation in the clinical trial setting as inhibitors of angiogenesis.

~~23 This process of the metastatic cascade was introduced in Chapter One and summarised in Figure 1.5: The Metastatic Cascade. 123~~

~~Chapter 2: Gene Expression Patterns in STS Progression~~

Figure 2.13 The MAPK signalling cascade. This is the integration point for a plethora of biochemical signals from growth factors, cytokines to physiologic stressors. It is involved in numerous processes such as cellular proliferation, cell survival, transcriptional regulation and development. The three main pathways that comprise this cascade are shown in the purple boxes on the left hand side. The classical pathway is the Sos-Ras- Raf pathway (top) activated by receptor tyrosine kinases. The GTP protein Ras activates Raf which then leads onto the series of phosphorylation and activation of the MAP kinases. This eventually promotes cell proliferation and survival. Cell survival may be mediatied by the anti- apoptotic bcl-2 gene. EGFR, which was overexpressed in the high grade and metastatic sarcoma cell lines in this study, activates this pathway. The JNK/p38 pathway (middle) is thought to be activated in response to stress, as well as via TGF! and IL1. TGFB1 and IL1B were overexpressed in both the high grade and metastatic cell lines in this study (circled). A number of MAP kinases downstream of these ligands were also overexpressed, namely, TAK1 (MAP3K7) , p38 delta (MAPK13/SAPK4) , HGK (MAP4K4) and MINK1. TGF! activated kinase, or TAK1 activates the MAPK8/JNK pathway, as does HGK/MAP4K4. MST4 (Mst3 and SOK1 related kinase) which activates ERK, was also upregulated. This figure was adapted from the Kyoto Encyclopaedia of Genes and Genomes or KEGG database (http://www.genome.jp/kegg/pathway.html)

~~124~~

~~Chapter 2: Gene Expression Patterns in STS Progression~~

The potential interaction between TGF! and integrins and MMPs via SMAD- independent pathways also warrants consideration as all of these factors are overexpressed in the high grade and metastatic cell lines compared to the low grade cell line, with some showing a clear progression in expression. In the following chapter, the expression of ITGA2 and ITGA6 is validated using qRT-PCR.

~~2.4.2.7 The Notch Pathway~~

NOTCH4 was overexpressed in the high grade and metastatic cell line in this study, as was a related receptor, DNER (Delta- and Notch-like EGF related receptor). The Notch gene, originally described almost 100 years ago after the ‘notches’ in the wings of a mutant fruitfly, was determined to be required for normal wing development. Since then, Notch has been identified as part of an ancient cell signalling system (Figure II.2, Appendix II) involved in cell fate determination and development (Esni, Ghosh et al. 2004; Grego-Bessa, Diez et al. 2004; Lai 2004). Both the ligands that bind to the transmembrane notch receptors (receptors Notch 1-4), Delta 1, Delta 3, Delta 4 and Jagged 1-2 and the receptors themselves, comprise a variable number of epidermal growth factor (EGF)-like repeats, which are considered to be essential components of signalling. The NOTCH4 protein contains 29 EGF-like repeats. In normal tissue, this pathway plays an essential role in regulating embryonic vascular, kidney morphogenesis and remodeling, as well as limb bud and pancreas development. Downstream signal transduction involves transcription factors such as the Hairy and Enhancer of Split families (HES), as well as the cell cycle regulator p21 and the MAPK signalling pathway. NOTCH1 overexpression has been noted in synovial sarcomas (Francis, Namlos et al. 2007). Aberrant NOTCH4 has also been reported in pancreatic cancer (Crnogorac- Jurcevic, Efthimiou et al. 2002), breast cancer (Politi, Feirt et al. 2004) and ovarian cancer (Donninger, Bonome et al. 2004). It is proposed that Notch prevents tumour cells from differentiating. Tumour suppressor effects have, however, also been described in animal studies. NOTCH 1-3 is involved in the normal differentiation of the epidermis. There is growing evidence that Notch promotes epithelial-mesenchymal transition (EMT) in tumour progression by increasing cell motility and decreasing polarity via Snail and Cadherins (Thiery 2002; Radtke and Raj 2003; Noseda, McLean et al. 2004;

~~125~~

~~Chapter 2: Gene Expression Patterns in STS Progression~~

~~Timmerman, Grego-Bessa et al. 2004). There may, in fact, be integration of TGF! and Notch signalling in this respect (Zavadil, Cermak et al. 2004).~~

Notch 1,2 and 4 is expressed in the osteosarcoma cell line SaOS-2 (Schnabel, Fichtel et al. 2002) and more recently in another study of soft tissue sarcomas (Yoon, Segal et al. 2006) but the role of Notch signalling in soft tissue sarcomas is yet to be clearly defined. It is possible that it plays a role in angiogenesis within these tumours as has been described in relation to other tumour types (Li and Harris 2005; Shi and Harris 2006).

~~2.4.3 SUMMARY~~

The results of this study concur with the growing body of knowledge in the field of oncology, that is, despite its complexities, repeated similarities are revealed in any study of tumor progression and metastasis. What can be appreciated is the likelihood of cross- talk, the integrated nature of the circuitry involved in tumour progression (Figure 2.14).

Vogelstein’s multi-step model of colorectal carcinogenesis (Vogelstein, Fearon et al. 1988) showed that a number of rate limiting events are required for malignant transformation. These “acquired capabilities” of the malignant cell (Hanahan and Weinberg 2000), essentially involve the evasion of the usual homeostatic checks placed on normal cells. Metastasis, as exemplified by the “metastatic cascade” (Cotran, Kumar et al. 1994) is a similar multi-hit process. The development of the metastatic subclone results cells capable of invasion by degradation of ECM, intravasation, tumour embolus formation, deposition at a distant site and neovascularisation. An alternate theory of metastasis, is that metastatic capability is present in all cells of the primary tumour, rather than as a late acquired mutation (Ramaswamy, Ross et al. 2003). This ability simply becomes apparent late in the stage of tumour development (Bernards 2003). In either case, the clues to metastasis are likely to be found by the examination of tumours at various stages of progression.

~~126~~

~~Chapter 2: Gene Expression Patterns in STS Progression~~

Figure 2.14 The integration of signalling pathways. Some of the components of this integrated circuit have been discussed in this section, including the growth factor tyrosine kinase receptor pathway that activates classical MAP kinase signalling. TGF! direct activation of MAP kinases, independent of Smad has also been suggested, in light of the findings of this gene expression study. Integrins and matrix metalloproteinases are also involved, particularly in relation to cell motility, invasion and neovascularisation. Cyclins B2 and D were some of the cell cycle regulators also shown to be upregulated in the higher grade cell lines, contributing to cell proliferation. This figure was reproduced from Hanahan and Weinberg’s discussion on “The Hallmarks of Cancer” (Hanahan and Weinberg 2000)

This gene expression study aimed to examine factors and pathways involved in tumour progression in soft tissue sarcomas, using sarcoma cell lines of increasing metastatic potential as a model. Functional groups of genes comparatively overexpressed in the high grade and metastatic cell lines have been identified, including • EGFR, its ligand, EREG and other genes with EGF-motifs • The TGF! and activin family • Members of the MAPK cascade • Integrins and MMPs • Cell cycle regulators and

~~127~~

~~Chapter 2: Gene Expression Patterns in STS Progression~~

~~• NDRG1.~~

EGFR was chosen for further study as differential expression had been demonstrated for factors at all stages of its pathway, from ligands and receptor to downstream signal transducers (MAPK cascade). A number of other molecular markers from the above categories were also selected for validation of the gene expression data. This was required to determine the utility of EGFR as a prognostic marker and its potential as a therapeutic target.

~~128~~

~~Chapter 3: Candidate Gene Validation~~

~~2"2~~

~~3. VALIDATION of TRANSCRIPT EXPRESSION of EGFR and OTHER CANDIDATE BIOMARKERS~~

~~129~~

~~Chapter 3: Candidate Gene Validation~~

~~3.1 INTRODUCTION~~

In the previous chapter, several members of the EGFR pathway were identified as being upregulated in the high grade and metastatic sarcoma cell lines, indicating a possible role for EGFR in sarcoma tumour progression and metastasis. Other differentially expressed genes were also discussed. Some of these candidate genes were selected for further validation and investigation.

Many sources of variation have been identified at each step of a microarray experiment, from inter-laboratory differences (Irizarry, Warren et al. 2005), choice of microarray platform (Rhodes, Barrette et al. 2002; Jarvinen, Hautaniemi et al. 2004; Bammler, Beyer et al. 2005; Larkin, Frank et al. 2005), data analysis algorithms (Michiels, Koscielny et al. 2005) to underestimation of actual mRNA fold changes (Yuen, Wurmbach et al. 2002). Validation is necessary in light of the above. Early publications of microarray work concentrated on the array data itself, and no additional validation findings were presented (Khan, Simon et al. 1998; Golub, Slonim et al. 1999; Alizadeh, Eisen et al. 2000; Perou, Sorlie et al. 2000). Similarly, more recent studies profiling sarcomas have tended to present array data only (Nielsen, West et al. 2002; Shmulevich, Hunt et al. 2002; Lee, John et al. 2003; Segal, Pavlidis et al. 2003; Lee, John et al. 2004), although some have gone on to publish validation studies at a later date (Nielsen, Hsu et al. 2003).

Verification of mRNA expression levels using independent methods on either the same samples as used for the microarrays or a different cohort of samples is increasingly becoming the norm. These methods of validation include Northern analysis, RT-PCR and quantitative RT-PCR in real time (qRT-PCR). Recent gene expression studies have employed a variety of these techniques as their validation experiments. Real time RT- PCR has been used in studies involving breast cancer progression (Ma, Salunga et al. 2003), hereditary breast cancers (Hedenfalk, Duggan et al. 2001), pancreatic intraepithelial neoplasia (Prasad, Biankin et al. 2005) as well as soft tissue sarcoma (Nagayama, Katagiri et al. 2002). The rationale for validation experiments and the selection of quantitative real time RT-PCR (qRT-PCR) for this purpose are presented in greater detail in Appendix III.

~~130~~

~~Chapter 3: Candidate Gene Validation~~

~~3.1.1 Advantages of Real Time PCR~~

PCR in real time has the advantage over conventional techniques such as Northern analysis and conventional RT-PCR of being able to assess mRNA expression of a greater number of candidate genes in a larger number of samples. In contrast to Northern analysis, PCR in real time avoids the use of radioactive reagents, is quicker and more expedient to perform and requires a smaller amount of starting material. One of the main advantages of qRT-PCR is that it enables quantitation. There is also the potential to multiplex the expression of several genes in one assay using different fluorescent labels.

~~3.1.2 Normalisation~~

Relative quantitation of the mRNA or cDNA copy number for a given gene requires it to be “normalised” to a control sample or a housekeeping gene (HKG) (Pfaffl 2001; Radonic, Thulke et al. 2004). The HKG chosen should be one that is not regulated or influenced by experimental procedures (Kaytan, Yaman et al. 2003). HKG or non- regulated genes such as GAPDH, ACTB (β-actin), 18S and B2M (β2 Microglobulin) have traditionally been used in qRT-PCR assays. In the present study, two HKGs, GAPDH and ACTB were assayed on each experiment. As the cell lines used in this study were all soft tissue sarcoma lines cultured under similar conditions, with no additional treatment and harvested at similar levels of confluence, it was assumed that the mRNA copy numbers for GAPDH and ACTB would be uniform.

~~3.1.3 Relative Quantitation~~

The comparative Ct method (Pfaffl 2001) normalises the expression of the target transcript against a housekeeping gene and compares the expression of the transcript in the sample to a control. For this method to be applied, the amplification efficiencies of the target and normaliser have to first be determined. The calculations involved are described in greater detail in the methods section (Section 3.2.6). The relative quantitation of gene expression can also be analysed with the same control sample and housekeeping genes using the Relative Expression Software Tool (REST 2005 v1.9.10) (Pfaffl, Horgan et al. 2002). Randomisation and bootstrapping techniques are applied and 50 000 iterations are carried out rather than traditional statistical tests. Primer and reaction efficiencies are taken into account. Where multiple reference genes are used, a

~~131~~

~~Chapter 3: Candidate Gene Validation~~

~~geometric mean is calculated and expressed as a “normalisation factor(Vandesompele, De Preter et al. 2002). The equation used is given in the methods section.~~

~~3.1.4 Candidate Genes Investigated~~

Some of the functional groups of genes and associated pathways over-expressed in the high grade and metastatic cell lines compared to the low grade sarcoma were considered in Chapter 2 (Section 2.4.4). These genes may be involved in tumour progression and metastasis and hence could serve as potential prognostic markers or therapeutic targets. A selection of these over-expressed genes was examined using qRT-PCR. In addition, a number of genes not found to be differentially expressed on the microarray analysis were also investigated, to confirm the validity of the results. In this chapter, the expression of EGFR, ACVR2B, TGFB1, MAP4K4, CCND1, ITGA2, ITGA6, NDRG1, HMGB1, and PDGFRA is determined using qRT-PCR. EGFR was most highly expressed in the metastatic cell line GCT and was selected. The next 4 of the above were all over-expressed in the high grade tumour cell line HT1080 compared to the low grade SW684 and were therefore selected as candidate genes for verification. The integrins ITGA2 and ITGA6 were highly expressed in the high grade cell line, with lower expression in the metastatic cell line GCT. The same pattern was observed for NDRG1. HMGB1 and PDGFRA, which were not statistically differentially expressed was included in the group of genes to be verified by qRT-PCR.

~~3.1.4.1 Epidermal growth factor receptor (EGFR)~~

~~The gene encoding EGFR is located on chromosome 7 and comprising 26 exons was described in detail in 1987 by Haley and colleagues (Haley, Whittle et al. 1987).~~

EGFR is a transmembrane tyrosine kinase receptor, to which EGFR as well as EGF-like molecules such as epiregulin can bind. Ligand binding results in dimerisation of the receptor, activating downstream signalling cascades.

~~132~~

~~Chapter 3: Candidate Gene Validation~~

Figure 3.1 Gene locus for EGFR The epidermal growth factor receptor has been mapped to the above locus, 7p11.2, as indicated by the single line on the chromosome map. Other genes mapped to this region include calmodulin 1 pseudogene2 (CALM1P2) and EGFR-coamplified and over-expressed protein (ECOP). This image was obtained from the Weizmann Institute of Science’s Gene Cards database (http://bioinfo1.weizmann.ac.il/genecards/index.shtml).

These include the MAPK, phosphatidylinositol 3-kinase (PI3K), phospholipase C-γ (PLCγ) and the JAK-STAT pathways (Jorissen, Walker et al. 2003). EGFR and epiregulin (EREG) were found to be over-expressed in the metastatic cell line GCT.

~~3.1.4.2 Activin Receptor Type 2B (ACVR2B)~~

The activin receptors are transmembrane proteins composed of a ligand-binding extracellular domain, a transmembrane domain and a cytoplasmic domain with serine/threonine activity. Type 2 receptors are required for binding ligands and for expression of type 1 receptors. Ligand binding results in Type 1 and 2 receptors forming a stable complex, and phosphorylation of Type 1 receptors by Type 2 receptors which triggers the signalling cascade described in Section 2.4.5.2 (Figure 2.12). The gene maps to Chromosome 3, at locus 3p22-p21.3, as indicated below.

Figure 3.2 Gene locus for Activin receptor Type 2B (ACVR2B) This is indicated by the single line on chromosome 3, at 3p22-p21.3. This image was obtained from the Weizmann Institute of Science’s Gene Cards database (http://bioinfo1.weizmann.ac.il/genecards/index.shtml).

Activins, in normal tissue, are known to exert inhibitory effects on cell proliferation, and enhance apoptosis (Chen, Lui et al. 2002). Aberrations in the activin signalling pathway have been reported in various malignancies, including breast cancer, prostate cancer and leukemia (Lappohn, Burger et al. 1989; Shin, Wang et al. 2003; Ball, Mellor

~~133~~

~~Chapter 3: Candidate Gene Validation~~

et al. 2004). It was suggested that Inhibin A (Activin) exerts an initial tumour suppressor effect in early stage prostate cancer and a subsequent tumour promoting effect in advanced stages (Ball, Mellor et al. 2004). A similar role may eventually be elucidated for the activins in sarcoma.

~~3.1.4.3 Transforming growth factor beta 1 (TGF! 1)~~

~~TGFB1 is known to map to chromosome 19 as shown in Figure 3.3.~~

Figure 3.3 Transforming growth factor beta 1 at 19q13.1 The gene locus is indicated by the single line on the chromosome map. Other genes mapped to this region include various zinc finger proteins, G-protein coupled receptors, an inhibitor of nuclear factor kappa light chain (NFκκκB1B), RASGRP4, a member of the family of guanine nucleotide exchange factors that assist Ras signalling by dissociating bound GDP and a subunit if platelet activating factor (PAFAH1B3), among others. This image was obtained from the Weizmann Institute of Science’s Gene Cards database (http://bioinfo1.weizmann.ac.il/genecards/index.shtml). TGF! 1 controls proliferation and differentiation in a variety of cell types. TGF! 1 signalling in cancer has been discussed in greater detail in Chapter 2 (Section 2.4.5.2), including the involvement of TGF! in both tumor suppression and cancer progression. The differential effects of TGF! 1 on epithelial and mesenchymal tissue have also been considered. In this study, TGFB1 was over-expressed in the high grade sarcoma cell line compared to the low grade on analysis of the microarray data.

~~3.1.4.4 Mitogen-activated protein kinase kinase kinase kinase 4 (MAP4K4)~~

MAP4K4 expression at mRNA transcript level is examined further in this chapter. As discussed in Chapter 2 (Section 2.4.5.5), the MAPK signalling cascade is the integration point for a plethora of biochemical signals from growth factors, including TGFB and EGFR. The protein encoded by MAP4K4 (also known as HGK) is a member of the serine/threonine protein kinase family and has been shown to specifically activate MAPK8/JNK and is thought to function through the MAP3K7-MAP2K4-MAP2K7 kinase cascade, and mediate the TNF-α signalling pathway.

~~134~~

~~Chapter 3: Candidate Gene Validation~~

Figure 3.4 Gene locus for MAP4K4 This has been mapped to the above locus, 2q11.2-q12, as indicated by the single line on the chromosome map. It is a serine/threonine kinase that may play a role in the response to environmental stress and cytokines such as TNF-alpha. This image was obtained from the Weizmann Institute of Science’s Gene Cards database (http://bioinfo1.weizmann.ac.il/genecards/index.shtml).

~~3.1.4.5 Cyclin D1 (CCND1)~~

This gene, located on 11q13 (Tsujimoto, Yunis et al. 1984), was initially cloned as a breakpoint rearrangement in parathyroid adenomas (PRAD1) (Arnold, Kim et al. 1989). CCND1 is now known to have both cyclin dependant kinase (CDK) mediated and CDK independent functions (Lamb and Ewen 2003; Ewen and Lamb 2004; Fu, Wang et al. 2004; Knudsen, Diehl et al. 2006). The CCND1-CDK4 complex is involved in progressing cells through the G1-S phase of the cell cycle (Fu, Wang et al. 2004). This function is promoted by growth factors such as EGF and IGF. The CDK-independent activities include modifying gene transcription of signal transducers and hormones such as STAT3, C/EBPβ, BETA2 and oestrogen receptor (ER) (Lamb and Ewen 2003; Ewen and Lamb 2004; Fu, Wang et al. 2004; Knudsen, Diehl et al. 2006). CCND1 is postulated to exert an inhibitory effect on STAT3.

Over-expression of CCND1 has been noted in a variety of epithelial cancers, including squamous cell cancer of the head and neck (Freier, Joos et al. 2003), colorectal cancer (Bala and Peltomaki 2001), bladder cancer (Wang, Habuchi et al. 2002), breast and oesophageal cancer (Knudsen, Diehl et al. 2006). In most cases, over-expression of CCND1 is secondary to induction by oncogenic signals such as β catenin, Ras and ErbB2. .

In this thesis, CCND1 was over-expressed in the high grade sarcoma cell line HT1080 compared to the low grade SW684 on gene expression array. CCND1 expression is examined further in this chapter by qRT-PCR.

~~135~~

~~Chapter 3: Candidate Gene Validation~~

~~3.1.4.6 N-myc downstream regulating gene 1 (NDRG1)~~

NDRG1 or DRG1 (differentiation-related gene 1) was first found to be induced during in vitro differentiation of colon cancer cell lines and cloned in 1997 (van Belzen, Dinjens et al. 1997). NDRG1 has been shown to be expressed in the epithelial cells of the small intestine, prostate, liver, lung, ovary and pancreas (Lachat, Shaw et al. 2002). It is known to be modulated by factors such as hypoxia (Park, Adams et al. 2000), androgens and p53 (Stein, Thomas et al. 2004).

NDRG1 over-expression has been investigated in epithelial cancers, as discussed in Chapter 2. However, its expression in soft tissue sarcomas is yet to be determined. NDRG1 over-expression in the high grade sarcoma cell line HT1080 is validated in this chapter.

~~3.1.4.7 Integrins alpha 2 and 6 (ITGA2 and ITGA6)~~

Integrins are heterodimeric transmembrane receptors comprising α and β subunits (van der Flier and Sonnenberg 2001). Integrins control attachment to the extracellular matrix (ECM) via ligands such as fibronectin and laminin. The integrin-ligand interaction results in changes to the actin cytoskeleton and additionally triggers signal transduction cascades that promote cell motility, proliferation and survival (Figure II.3, Appendix II) (van der Flier and Sonnenberg 2001; Jin and Varner 2004).

Integrins are also thought to interact with receptor kinases such as EGFR and ErbB2 (Falcioni, Antonini et al. 1997; Moro, Venturino et al. 1998; Gambaletta, Marchetti et al. 2000; Yu, Miyamoto et al. 2000). Association of integrin heterodimer α2β1 with EGFR results in activation of the MAPK pathway (Moro, Venturino et al. 1998). Integrins have also been linked to tumour progression and metastasis. Cellular migration can be mediated by integrin α6β1 and α6β4 or α2β1 and collagen, among others (Friedl and Wolf 2003). Activation of focal adhesion kinases (FAK) occurs. EGF and IGF1 are some of the factors which promote cell migration.

~~This gene expression study showed an increased expression of integrins α2 and α6 in the aggressive high grade cell line HT1080. Validation of this finding is carried out using qRT-PCR.~~

~~136~~

~~Chapter 3: Candidate Gene Validation~~

~~3.1.4.8 High mobility group box 1 (HMGB1)~~

HMGB1 may be passively released from necrotic cells or actively secreted as an immune response (Erlandsson Harris and Andersson 2004; Schlueter, Weber et al. 2005; Ulloa and Messmer 2006). In its capacity as an inflammatory mediator, it promotes neutrophil chemotaxis and vascular permeability in response to tissue injury, sepsis or ischaemia. The MAPK signalling cascade is activated via p38MAPK, p44/42MAPK and SAPK/JNK. STAT1 and STAT3 are also known to be activated in macrophages (Lotze and Tracey 2005).

HMGB1 protein binds DNA, acting as a transcriptional regulator, modulating the expression of many genes (Muller, Scaffidi et al. 2001; Lotze and Tracey 2005; Ulloa and Messmer 2006). Genes known to be involved in malignancy such as E selectin and BRCA1 are regulated by HMGB1. Over-expression of HMGB1 has been reported in pancreatic, prostate and breast cancer (Tarbe, Evtimova et al. 2001; Brezniceanu, Volp et al. 2003; Leman, Madigan et al. 2003). It was also over-expressed in KIT mutation positive gastrointestinal stromal tumours (GISTs) (Choi, Kim et al. 2003). Furthermore, it has been purported to be involved in tumour metastasis (Nestl, Von Stein et al. 2001; Evans, Lennard et al. 2004).

There was no significant difference in HMGB1 expression of in the microarray analysis in the present study. However HMGB1 expression was further examined in this chapter using real time RT-PCR as it has been reported to be over-expressed in GISTs. HMGB1 expression was also evaluated in the newly developed GIST cell line GIST-M and the leiomyosarcoma cell line LMS-LFS as detailed in Chapters 6 and 7.

~~3.1.4.9 Platelet derived growth factor receptor A (PDGFRA)~~

This gene encoding a tyrosine kinase receptor is located on chromosome 4q11-q12, in the same region as the KIT proto-oncogene (Gronwald, Adler et al. 1990). Ligand binding to the growth factor results in homo- or heterodimerisation (with PDGFRB) and activation. PDGFRA is known to stimulate the growth of cells of mesenchymal origin. It plays a significant role in angiogenesis. PDGFRA has been extensively investigated in the context of GISTs (Heinrich, Corless et al. 2003b; Corless, Fletcher et al. 2004; Subramanian, West et al. 2004; Penzel, Aulmann et al. 2005). Most recently, a gene

~~137~~

~~Chapter 3: Candidate Gene Validation~~

expression study of 38 STS reported increased expression of PDGFRA compared to normal tissue (Yoon, Segal et al. 2006). PDGFRA was not significantly differentially expressed on microarray analysis in the present study, and this was further examined in this chapter using real time RT-PCR. Given the presence of PDGFRA mutations or over-expression in KIT-negative GISTs, its expression was also evaluated in the newly developed GIST cell line GIST-M and the leiomyosarcoma cell line LMS-LFS as detailed in Chapter 5.

~~3.2 METHODS~~

~~3.2.1 Cell Culture~~

The same cell lines as were used in the microarray experiments, MRC5, SW684, HT1080 and GCT were cultured as described previously in Chapter 2, doubling time determined and harvested at 70 – 80 % confluence for RNA extraction. Two additional sarcoma cell lines, SW872, a liposarcoma and SW982, a synovial sarcoma were included in selected experiments for comparison. These last two were not used to assess differential expression with respect to tumour grade as the grade of tumour from which they were derived, is not known.

~~3.2.2 RNA Extraction and cDNA Synthesis~~

RNA was isolated using TRIzol® (Invitrogen, Life Technologies. Carlsbad, CA) as described previously in Chapter 2 (detailed in Appendix I). The quality and concentration of the RNA was confirmed using both the Agilent 2100 BioAnalyzer (Agilent Technologies, Palo Alto, CA) and the NanoDrop ND-1000 (NanoDrop Technologies, Delaware, USA). cDNA was synthesised using a first strand cDNA synthesis system for RT-PCR (Marligen Biosciences, Ijamsville, MD) comprising oligo(dT), dNTPs, RNase H + and reverse transcriptase according to the manufacturers instructions, using 1 μg of RNA per 20 μl reaction. The concentration and quality of the resultant cDNA was assessed on the NanoDrop ND-1000, ensuring equivalent concentrations for all samples at the start of the qRT-PCR.

~~138~~

~~Chapter 3: Candidate Gene Validation~~

~~3.2.3 Primer Design~~

Primers were designed using Primer3 software developed by S. Rozen and H. J. Skaletsky http://www-genome.wi.mit.edu/genome_software/other/primer3. The criteria used included primer length (18-22 bp), amplicon length (130-300 bp), GC content (50 - 55 %), primer melting temperature (Tm) (~ 60 °C) and avoidance of self complementarity. The criteria were determined so as to reduce the likelihood of primer dimers and non specific binding to target transcript (Brownie, Shawcross et al. 1997; Gorelenkov, Antipov et al. 2001; Vandesompele, De Paepe et al. 2001). All primers were designed to have approximately the same Tm such that multiple target genes could be assessed at the same time, using the same cycling parameters. The primer sequences chosen were analysed for homology using National Centre for Biotechnology Information (NCBI) BLAST database http://www.ncbi.nlm.nih.gov/BLAST/. Primers were obtained from Sigma Genosys, Castle Hill, Australia and resuspended with ddH2O at 100 ng/ μl concentration. The primer sequences and amplicon size for the candidate and housekeeping genes are shown in Table 3.1.

~~3.2.4 Real time RT-PCR (qRTPCR)~~

~~The target genes were quantified by qRT-PCR (ABI PRISM® 7700 Sequence Detection System, PE Applied Biosystems or Rotor-Gene™ 3000, Corbett Life Science) using~~

Platinum® SYBR® Green qPCR SuperMix-UDG (2X) (Invitrogen, Life Technologies). This system consists of Platinum Taq DNA polymerase which employs the “hot start” mechanism where the polymerase is not activated until the denaturing step of PCR cycling. Competing side reactions such as mispriming and primer dimerisation can alter the sensitivity of the PCR reaction (Bustin 2002). Heat-activated polymerase theoretically avoids mispriming and improves the yield for the specific product. SYBR® Green fluorescent dye binds to the minor groove of DNA, emitting a signal that is proportional to the dsDNA concentration. 1 μl of cDNA template was used in each 25 μl reaction.

The cycling parameters were as follows: 50 °C for 2 min, 95 °C for 2 min, then 35 cycles of 95 °C for 15 s and 60 °C for 30 s. For ITGA6, the annealing temperature was optimised at 58 °C. A melt curve analysis was carried out of the amplified products,

~~139~~

~~Chapter 3: Candidate Gene Validation~~

increasing the temperature from 60 °C to 95 °C over a ramp time of 20 min to ensure that the fluorescence of the specific product was being recorded. The peak of the curve represents the temperature at which 50 % of the amplicon melts into single stranded DNA. The amount of fluorescence reflected the amount of transcript amplified in an indirect manner. GAPDH and ACTB were both used as internal controls and included on every run. Each target gene for each template cDNA sample was run in duplicate for each run and each experiment was repeated at least three times.

Table 3.1 Primers used in real time RT-PCR &#,#< +#&#, ,) .0'+#0#/3#,!#13+ 8+.*'!-, ZZ <RRVZUV "SW′V***3**33**3&33**3*3383VU′ XUTV°* TSW (SW′V8&8*8***8*88&8*8&8&3*8VU′ XUTZ°* $0#X <RRSSRX "SW′V*8&*3&&83&3&38&&3&888&VU′ XUTW°* SYZ (SW′V*3**8&33*8&8&3***8333*VU′ XUTZ°* /W <RRRXXR "SW′V**8&8&8&&3388&&&8&&8&3VU′ XUTX°* SZZ (SW′V*383&338***8&&*3&&3*3*VU′ XUTX°* /# <RRWTTZ "SW′V&*38&338&&8&***8**3333VU′ XUTV°* TVY (SW′V*3&38&&&&*3*3&8*3&83*3VU′ XUTX°* $$2W <RRSYWZ "SW′V&3&*3&*&88&3&&888**VU′ XSTX°* SYV (SW′V83**8&&3&&*&8*&83*3VU′ XTTY°* X <RRTTRU "SW′V&**3*&&*33&83**3*8*VU′ XTTW°* SRS (SW′V8***&33&3&38838*3&833**VU′ XRTR°* \ <RRRTSR "SW′V33&&8&*3333&3&83&&&*VU′ XRT[°* SVU (SW′V*8*&88*883***333**8&33VU′ XRTZ°* 2#W <RRXR[X "SW′V8*88*****3*33*88*38*&VU′ XSTV°* S[S (SW′V&**883883&*3333*8&***88VU′ XTTT°* W <RRTSTZ "SW′V8&3&*3*8&8&8&&3&&88&8*VU′ XUTZ°* S[U (SW′V3*8&8&*8&88&8&&88&88&&VU′ XUT[°* /# <RRXTRX "SW′V888*8&&8&&8&8*3&*88&8&VU′ XUTY°* SWV (SW′V&88&8*8&**388&8**8&&88VU′ XUTV°* $ *RSVZXS "SW′V*8**8*8**33*38*883&8&*VU′ XUTV°* SXU (SW′V838&*8*8&**3&&838&*88*VU′ XVTT°* <RRTRVX "SW′V&&3&&3*3**3*3&8*33*88*VU′ XUT[°* TST (SW′V*3*33**3*33&3&*3*33&*3VU′ XUT[°*

~~140~~

~~Chapter 3: Candidate Gene Validation~~

~~3.2.5 Agarose Gel Electrophoresis~~

Samples were also assessed by electrophoresis to confirm the amplicon sizes. 5 )l of the total reaction volume was combined with 1 )l gel loading dye (Ambion, Austin, TX). Samples were separated on a 3 % (w/v) agarose gel that was subsequently stained in 20 )g/ml ethidium bromide, in TAE buffer (40 mM Tris-acetate, 1 mM EDTA) using 80 V. A 100 bp 1 Kb DNA ladder (Hyperladder IV, Bioline, London, UK) was used to assess the bands obtained. Gels were then visualised and photographed on a Gel Doc system (Biorad Laboratories, Hercules, CA).

~~3.2.6 Relative quantitation~~

There are two methods of relative quantitation, the standard curve method and the comparative Ct method. The latter method was initially adopted as it allows for increased throughput, as wells on every 96-well plate no longer need to be used for the samples required to generate the standard curves. It is therefore also the more economic method of the two. The “housekeeping” (HKG) or non-regulated genes GAPDH and ACTB were used normalisation. The calibrator or reference cDNA in this case, was that of the normal fibroblast cell line MRC5, as for the microarray experiments.

~~The comparative Ct method uses the mathematical formula described by Michael Pfaffl (Pfaffl 2001) as TΔΔΔΔΔΔ2~~

Where ΔΔCt = ΔCt(sample) - ΔCt(reference) ΔCt sample= Ct(sample)- Ct(HKG) ΔCt(reference) =Ct(calibrator)- Ct(HKG) For the above formula to be applied, amplification efficiencies of the target and the endogenous reference gene or HKG must be approximately equal. Serial dilutions of pooled cDNA from all cell lines were used to determine the efficiency for each target gene and housekeeping gene. The efficiency of the reaction was calculated by plotting the Log of the cDNA concentrations against the mean threshold cycle. The slope of the resulting line of best fit was then used in the formula given below to calculate PCR efficiency (E) (Pfaffl 2001).

~~141~~

~~Chapter 3: Candidate Gene Validation~~

~~E = 10^(-1/Slope)~~

~~If E target gene = E housekeeping gene = 2, then the above formula for relative quantitation could be applied.~~

Subsequently, the Relative Expression Software Tool (REST© 2005) was applied to all real time RT-PCR gene expression data to recalculate relative expression ratios (Pfaffl, Horgan et al. 2002). The advantage of using this software tool was that the calculation can take into consideration more than one reference or housekeeping gene at the same time. Both reference genes used in the experiments, GAPDH and ACTB, could thus be incorporated into the one equation. This mathematical model, using a Pair Wise Fixed Reallocation Randomisation Test©, carries out 50 000 iterations to arrive at the expression ratio from the following equation:

~~ΔCPtarget(control-sample) ΔCPref(control-sample) Ratio = (Etarget) / (Ereference)~~

Where E = PCR Efficiency, as given by the equation above ΔCP = crossing point difference of a sample versus control. The threshold cycle is used to determine ΔCP. Group means for target genes and reference genes are calculated, with the normalised ratios presented for comparison.

Finally, an analysis of variance (ANOVA) was carried out on the log2 value of the expression ratios, presenting an F ratio. The F-Test is a parametric test which is based on certain assumptions of normal distribution of data points. Gene expression abundances are rarely normal, but are frequently log-normal. The F ratio is taken as the ratio of Mean Square Between (samples) to Mean Square Error (F = MSB/MSE). Mean Square Between (MSB) is based on the variance of the sample means. If the null hypothesis is true, then the F ratio should be approximately one since MSB and MSE should be about the same. If the ratio is much larger than one, then it is likely that MSB is estimating a larger quantity than is MSE and that the null hypothesis is false.

~~The p value provides a measure of significance and comparison of the log2 transformed expression ratios provides a measure of the biological impact.~~

~~142~~

~~Chapter 3: Candidate Gene Validation~~

~~3.3 RESULTS~~

~~3.3.1 Amplification Plots~~

A representative series of amplification curves from one real time PCR experiment is shown in Figures 3.5. The replicates group together tightly and most of the products enter the exponential phase of amplification much earlier than the non template controls (NTC).

~~3.3.2 Assessment of PCR Product Specificity~~

~~3.3.2.1 Melt Curves~~

The corresponding melt curves from the same experiment in Figure 3.5 are shown in Figures 3.6. Melt curves for two target genes examined on the Rotor-Gene® 3000 Sequence Detection System (Corbett Life Science, Sydney, Australia) are shown in Figure 3.8. The melt curves served as a measure of product specificity. The specific amplicons have a higher melting temperature that the primers or the primer dimers (PDs) that form in the reaction where there is no template cDNA. PDs can also occur in cases where the transcript of interest is present in great abundance, such as with GAPDH. In these cases, increasing the amount of template cDNA can reduce the formation of PDs.

~~3.3.2.2 Agarose gel electrophoresis~~

For selected plate runs, 5 μl of each amplicon was electrophoresed on 3 % (w/v) agarose gels stained with ethidium bromide as described. This was done as an additional confirmation of product specificity. The amplicon length was also determined in this manner. A representative selection of the gels is shown in Figure 3.7. Two additional sarcoma cell lines, SW872 and SW982, a liposarcoma and synovial sarcoma respectively, are included on the gels. They were not however used in either the microarray analysis or qRT-PCR relative quantitation as the grade of tumour from which they were derived is not known.

~~3.3.3 The housekeeping genes (HKG)~~

~~The expression of the HKG, GAPDH was consistent across all the cell lines assessed in this study. The Ct values (=Cp as mentioned in the above Pfaffl calculation) for~~

~~143~~

~~Chapter 3: Candidate Gene Validation~~

GAPDH ranged between 14.09 and 14.61 on repeated runs. The expression of ACTB, however, showed a greater variation, with Ct values between 15.2 and 16.5. The expression of these HKG in the new cell line LMS-LFS, characterised in Chapter 5 also fell within the above ranges. The new GIST cell line, characterised in Chapter 5, had Ct values of 16.3 for GAPDH and 15.6 for ACTB, highlighting a potential issue with the relative quantification method for the latter cell line if the same normal fibroblast cell line MRC5 was to be used as the control with GAPDH as the HKG. This could potentially affect the values after normalisation. However, when an ANOVA was carried out for these genes, it was found not to be statistically differentially expressed across the sarcoma cell lines.

~~3.3.4 Amplification Efficiencies~~

A dilution series of pooled cDNA from all the cell lines was used for this experiment (Figure 3.8 and in Appendix III). A plot of the Mean Ct against the Log cDNA concentration was plotted (Figure 3.9 and Figures III.2 – III.3, Appendix III). The slope of the line of best fit was derived from this. As the gradient for each target gene and reference gene was close to 2, implying equivalent PCR efficiencies, the comparative formula was used for relative quantitation. This concurs with the optimal efficiency of 2 as described by Pfaffl (Pfaffl 2001). The amplification efficiency for each gene was also incorporated into the REST© 2005 software tool.

~~144~~

~~Chapter 3: Candidate Gene Validation~~

~~&0#1&-*" (~~

Figure 3.5 Amplification plots Examples of amplification plots as seen on the (a)ABI PRISM® 7700 Sequence Detection System (PE Applied Biosystems) and (b) Rotor-Gene® 3000 Sequence Detection System (Corbett Life Science, Sydney, Australia) for multiple target genes, housekeeping genes (HKG) and non template controls (NTCs). Each curve represents amplification of a product, as measured by the amount of SYBR® green fluorescence. The curves are colour coded for the gene of interest. The threshold (arrowed) is taken during the exponential phase of amplification, where the reaction is at its most efficient. The point at which this line crosses the respective curves determines the threshold cycle, Ct (circled). The lower the Ct value, the earlier the product amplification reaches its log phase, reflecting a higher concentration of the target transcript. Primer dimer amplification occurring where there is no template cDNA (NTC) have much higher Ct values.

~~Linear Scale: CCND1, GAPDH and ACTB~~

~~Log Scale: NDRG1, CCND1, GAPDH and ACTB~~

~~145~~

~~Chapter 3: Candidate Gene Validation~~

~~Target Genes GAPDH ACTB NTC Primer dimers~~

Figure 3.6 Product Specificity Melt curves obtained for target genes are shown in these plots. Once the real time RT-PCR reaction was complete, a melt analysis was carried out for each experiment to assess product specificity. The x-axis is the temperature range (60 οοοC – 95 οοοC) over which the melt analysis is carried out. The y- axis refers to the SYBR® green fluorescence recorded for each amplicon as it melts to ssDNA. (a)The expression of target genes ACVR2B, MAP4K4, HMGB1 and HKGs GAPDH and ACTB and (b) ITGA2 and ITGA6 were assayed for the same cell lines of increasing metastatic potential as were used in the microarray experiments (MRC5, SW684, HT1080 and GCT), as well as for the two primary cultures (GIST and LMS-LFS). Primer dimers that formed in the non template control (NTC) reactions melted at lower temperatures than the specific target and housekeeping amplicons. ITGA2 Sarcoma Cell Lines

~~ITGA6 Sarcoma Cell Lines~~

~~NTC~~

~~146~~

~~Chapter 3: Candidate Gene Validation~~

~~STUVWXY 200bp 178bp 100bp ACVR2B STUVWXY 200bp 215bp MAP4K4 100bp ! STUVWXY 200bp 247bp 100bp EGFR~~

Figure 3.7 Confirmation of Product Specificity The size of the amplicons was also confirmed for each target gene by electrophoresing on 3 % agarose gels stained with ethidium bromide. A selection of these is shown above, for the target amplicons (a) ACVR2B, (b) MAP4K4 and (c) EGFR. Lane 1 shows the 100 bp 1kb DNA ladder. Lane 2 is the normal fibroblast cell line MRC5, Lane 3 is the low grade fibrosarcoma cell line SW684, Lane 4 is the high grade HT1080 cell line and the metastatic cell line GCT is in Lane 5. Two other sarcoma cell lines, SW872 (liposarcoma) and SW982 (synovial sarcoma) are also included on this gel in Lanes 6 and 7 respectively. Recent microarray studies have reported increased expression of EGFR in synovial sarcomas.

~~147~~

~~Chapter 3: Candidate Gene Validation~~

~~EGFR Dilution Series Amplification Curves~~

~~NTC~~

~~Amplicons~~

~~NTC~~

Figure 3.8 PCR Dilution Series and Melt Curves for EGFR The amplification curves for the dilution series of pooled cDNA (Neat, 1:10, 1:50, 1:100, 1:500, 1:1000, each in duplicate) are shown in graph (a), with product amplification commencing at progressively later cycle numbers (x-axis) the more dilute the template. This dilution series is used to calculate the efficiency of the reaction, by plotting a graph of the threshold cycle, Ct, against the Log of the concentration of cDNA, as shown in Figure 3.12. The corresponding melt curves are shown in (b). Melting occurs at the same temperature but the peaks are lower, indicating less total product. NTC refers to the non template control.

~~148~~

~~Chapter 3: Candidate Gene Validation~~

~~PCR Efficiency: EGFR~~

~~25 Mean Ct. 20~~

~~15 y = -3.3315x + 29.845 2 10 R = 0.991~~

~~0 -0.500.511.52 2.5 3 3.5 Log (cDNA Conc. ng/ul)~~

~~Mean Ct (EGFR) Linear (Mean Ct (EGFR))~~

~~PCR Efficiency: MAP4K4~~

~~25 Mean Ct. 20~~

~~15 y = -3.2316x + 28.729 10 R2 = 0.9985~~

~~0 -0.500.511.522.53 3.5 Log (cDNA Conc. ng/ul)~~

Mean Ct (MAP4K4) Linear (Mean Ct (MAP4K4)) Figure 3.9 PCR Efficiency plots of (a) EGFR and (b) MAP4K4 This plot of the Mean Ct values against the Log (cDNA template concentration) is used to determine whether the relative quantitation method as described by Pfaffl can be utilised for the given target genes and chosen housekeeping gene (GAPDH). The slope of each line, as calculated from the line of best fit, is used to calculate the efficiency (E) of the amplification, using the formula (-1/Slope) E = 10^ . If the efficiency E target gene = E housekeeping gene = 2, then using the relative quantitation method, as given by Michael Pfaffl’s formula, 2^-(ΔΔΔΔΔΔCt), is justified. From the above examples, where a dilution series of pooled cDNA from all the cell lines was used, EEGFR = 2.04 and EMAP4K4 = 2.00. Similar values were obtained for other target genes. The Pfaffl formula was therefore used for subsequent relative quantitation.

~~149~~

~~Chapter 3: Candidate Gene Validation~~

~~3.3.5 Relative quantitation~~

The normalised log-transformed expression ratios for the transcripts investigated are presented graphically in Figures 3.10. The statistical analysis for assessing significant differential expression is shown in tabulated form in Tables 3.2 and III.1 (Appendix III).

~~3.3.5.1 EGFR~~

EGFR (previously ERBB), which has been shown in an earlier study to be in a gene cluster defining synovial sarcomas (Allander, Illei et al. 2002), was found in the present gene expression study, to be increasing in expression across the tumour grades. These results were borne out by the qRT-PCR validation of the mRNA expression levels in the three cell lines, with level of expression increasing across the cell lines of increasing metastatic potential. This differential expression was statistically significant, showing a high F ratio of 31.97 and a p value of 0.001.

~~3.3.5.2 ITGA2 and ITGA6~~

The gene expression arrays had found statistically significant differential expression across the 3 cell lines, with an F statistic 8.267 and 7.295 for ITGA2 and ITGA6 respectively (Appendix II, Table II.8). In particular, there was much higher expression of these two integrins in the high grade cell line, HT1080. The normalised expression ratios on qRT-PCR concurred with the above findings, with the highest expression seen in the high grade cell line HT1080.

~~On statistical analysis, the F ratios for ITGA2 and ITGA6 were 14.8 and 169.2 respectively and the p values were significant (Table 3.2). These results therefore validate the microarray findings.~~

~~3.3.5.3 NDRG1~~

Differential expression of N-myc downstream regulating gene 1 was also confirmed on qRT-PCR. In the microarray study, the greatest difference in expression was shown to be between the low and high grade tumour cell lines. The gene was represented twice on the human 19K oligonucleotide array under the two gene IDs AF039944 and NM_006096. The F statistic, derived for the 3 group comparison was 10.635 and 8.134 respectively for the two gene IDs and the ΔM, 3.26 and 3.50 respectively on the two

~~150~~

~~Chapter 3: Candidate Gene Validation~~

group analysis. In qRT-PCR, NDRG1 expression was greatest in HT1080, in keeping with the microarray findings, followed by GCT. The low grade cell line manifested the lowest expression of this transcript. The differential expression was highly statistically significant, with an F ratio of 49.232 and a p value of 0.005.

~~3.3.5.4 MAP4K4~~

There was a higher expression of MAP4K4 mRNA in HT1080 than SW684 (Figure 3.10), in agreement with the microarray findings. The expression profile across the grades varied slightly when ACTB was used in the calculations as the only HKG with ΔΔ the 2^- Ct formula. This is likely to be a reflection of the Ct value for ACTB in MRC5 being lower than that for any of the tumour cell lines. In other words ACTB is more abundantly expressed in MRC5. However, on statistical analysis, there was no significant difference in expression of either reference gene across the all the cell lines studied. Thus, both HKG were used in the REST© 2005 software program, deriving a geometric mean and the expression ratio derived by this method subjected to further statistical analysis. When the log2 value of the expression ratios derived in REST© 2005 was analysed for statistical significance, the F ratio was found to be 7.874, with a p value of 0.064.

~~3.3.5.5 ACVR2B expression~~

The results of the relative quantification (Figure 3.10) validated the oligonucleotide array findings, in that the mRNA expression was greater in the high grade cell line HT1080 than the low grade cell line SW684. Interestingly, the expression seen in the metastatic cell line GCT was higher still, indicating that this gene may be involved in tumour progression as well as metastasis. On ANOVA with Bonferroni correction, the F ratio was 5.409 and the difference was statistically significant with a p value of 0.029. The greatest difference is between the low grade and the metastatic cell line (Table III.1, Appendix III).

~~3.3.5.6 TGFB1~~

TGFB1 mRNA expression (Figure 3.10) also showed an increase from low to high grade but not for the metastatic cell line. The gene had been found to be significantly differentially expressed between low and high grade cell lines on the expression arrays.

~~151~~

~~Chapter 3: Candidate Gene Validation~~

Another related gene TGFBI (transforming growth factor beta –induced) (Appendix II, Tables II.2 and II.3) was statistically differentially expressed, although its expression was lower in the metastatic cell line, GCT. It seems therefore that members of the TGFB family are dysregulated in sarcomas and that some, if not all, may well be involved in tumour progression and metastasis.

The log transformed expression ratios calculated in REST© 2005 for TGFB1 were analysed for statistical significance using ANOVA. The F ratio was found to be 0.637, with a p value of 0.59. In other words the differential expression of TGB1 in the 3 sarcoma cell lines was not statistically significant.

~~3.3.5.7 HMGB1 and PDGFRA~~

There was no clear progression or differences seen in the expression patterns for HMGB1 (Figure 3.10) on either the real time RT-PCR results presented in this chapter or in the statistical analysis of the microarray data. HMGB1 was over expressed compared to MRC5 but showed no relationship to tumour grade. This concurs with the gene expression study. PDGFRA was very poorly expressed by all three of the tumour cell lines, relative to the normal fibroblast cell line MRC5. The differential expression on real time RT-PCR was not found to be significant on statistical analysis.

~~3.3.5.8 CCND1~~

Cyclin D1 had been found in Chapter 2 to be differentially expressed between the low and high grade cell line on Bioconductor R analysis (Appendix II, Table II.2), with a ΔM=2.23, indicating a greater than four-fold difference in expression. It was not, however, found to be statistically differentially expressed across all 3 cell lines.

The CCND1 transcripts, when assessed using real time RT-PCR, were generally underexpressed in the sarcoma cell lines, compared to the control cell line MRC5. It was, however, most underexpressed in the low grade cell line SW684. On statistical analysis, the F ratio was 16.819, with a p value of 0.001. This difference was significant both between the low and high grade cell lines, as well as between the low grade and metastatic cell line (Table III.1, Appendix III).

~~152~~

~~Chapter 3: Candidate Gene Validation~~

Table 3.2 Statistical Analysis of Differential Gene Expression 0 ,1!0'.2 0 2'- .4 *3# "\ SX[TTUW RTRRS $%&W V[TTUT RTRRW ( & UST[YU RTRRS ))$%W SXTZS[ RTRRS "X SVTZTY RTRRW +Z-Z YTZYV RTRXV +% & XTXZV RTRYZ ).&X/ WTVR[ RTRT[ /W RTXUY RTWZZ /W RTSRX RT[RU � 524 *3#1- 2 ',#"$-02&#U1 0!-+ !#***',#1 ,"2&#!-,20-*!#***',# 5#0#,-0+ *'1#"2-.0-4'"##6.0#11'-,0 2'-131',%2&# TRRW.0-%0 +T&#

#6.0#11'-,0 2'-12&31"#0'4#"5#0#*-%T20 ,1$-0+#" ,"13 (#!2#"2- ,"( +" 5'2& .-12&-!/-,$#00-,'!-00#!2'-,T &# 0 2'- ," . 4 *3# - 2 ',#" $0-+ 2&'1 , *71'1 0#.0#1#,2#" -4#$-0# !&! ,"'" 2#%#,#',4#12'% 2#"T",0 2'--$S 5-3*" #- 2 ',#"'$2 0' ,!#5'2&',# !&!#***',#5 12 +# 12&# 4 0' ,!# #25##,2&#!#***',#1T&#&'%&#0 2'-Q2&#%0# 2#02&#"'$$#0#,2' * #6.0#11'-,T

~~153~~

~~Chapter 3: Candidate Gene Validation~~

~~Gene Expression in Sarcoma Cell Lines~~

~~0 CCND1 ITGA2 ITGA6 NDRG1 EGFR ACVR2B TGFB1 MAP4K4 HMGB1 PDGFRA -2~~

~~-4 of Expression Ratio 2 -6~~

~~Log -8~~

~~-10 Gene of Interest~~

~~SW684 HT1080 GCT~~

Figure 3.10 Relative quantitation of gene expression data on real-time RT-PCR. The y-axis represents represent the logarithmic values of the expression ratios for the candidate genes. The initial expression ratios were derived by entering the raw Ct values into the REST©©© 2005 program, as described in the text. The error bars depicted are derived from the standard error of the mean of the log-transformed expression ratios. Genes of interest are shown on the x- axis. The normal fibroblast cell line MRC5 was used as the control gene for the calculations. The housekeeping genes (HKG) GAPDH and βββ-actin were used to normalise the expression ratios by calculating a geometric mean of their Ct values. The log transformed values of the expression ratios were analysed for statistical significance. The results of this analysis are presented in Table 3.2.

~~154~~

~~Chapter 3: Candidate Gene Validation~~

~~3.4 DISCUSSION~~

The aim of this chapter was to validate the differential expression of EGFR and a number of other transcripts. Real time RT-PCR was employed as a sensitive approach to evaluate relative expression of target transcripts in the same cell lines as in the microarray experiment. To this end, the normal fibroblast cell line, MRC5 was again used as the common reference to compare the sarcoma cell lines of increasing metastatic potential. Further detail on the housekeeping/reference gene selection is provided in Appendix III.

~~3.4.1 Candidate Gene Validation~~

~~3.4.1.1 EGFR~~

Both the gene expression arrays and real time RT-PCR experiments confirmed increasing expression of EGFR with increasing tumour grade. In the latter, differential expression was highly significant, with a p value of 0.001.

EGFR and the related HER2/ErbB2 has been extensively studied in epithelial malignancies, with over-expression demonstrated in breast, ovarian, lung and pancreatic cancer and glioblastoma (Torring, Jorgensen et al. 2000; Albanell, Rojo et al. 2002; Hirsch, Varella-Garcia et al. 2003; Mukohara, Kudoh et al. 2003; Thomas, Chouinard et al. 2003). As discussed in Chapter 2 (Section 2.4.4.3.2), EGFR expression has also recently been reported in STS, particularly in synovial sarcoma and the paediatric sarcoma subtypes of alveolar and embryonal rhabdomyosarcoma (Nielsen, Hsu et al. 2003; Baird, Davis et al. 2005; Ganti, Skapek et al. 2006).

These findings clearly have implications for targeted therapies. Moreover, as Phase I and II trials of various EGFR inhibitors are already in progress (Albanell, Rojo et al. 2002; Daneshmand, Parolin et al. 2003; Baselga, Albanell et al. 2005; Bates and Fojo 2005), the prospect of trialling these targeted therapies in STS is an attractive one.

~~Prior to considering the use of targeted therapies directed against EGFR, its protein expression in a variety of STS should be examined. This issue is dealt with in Chapter~~

~~155~~

~~Chapter 3: Candidate Gene Validation~~

~~4, where a tissue microarray of paraffin-embedded STS specimens is constructed. EGFR expression and that of some of its downstream signalling molecules is evaluated.~~

~~3.4.1.2 Integrins~~

The integrins ITGA2 and ITGA6 were most highly over-expressed in the high grade cell line HT1080, with expression decreasing in the metastatic cell line GCT. These genes were however under-expressed in the low grade cell line SW684. Invasion and metastasis required cell migration, ECM degradation, intravasation into blood vessels or lymphatics and extravasation at a distant site. As discussed in the introductory sections of this chapter, this cellular migration can be mediated by integrin α6β1 or α6β4 with laminin in the ECM or α2β1 and collagen, among others (Friedl and Wolf 2003). It seems intuitive, therefore, that the integrins alpha-2 and alpha-6 would be greatly increased in the high grade cell line, which has the potential to invade and metastasise. It should also be noted that the patient from whom the aggressive cell line HT1080 was derived, went on to develop metastases. Once metastasis has already occurred, as in the metastatic cell line GCT, the integrins may no longer need to be expressed at very high levels.

Integrin alpha-6 expression has been noted in other epithelial cancers, such as oral squamous cell cancer and breast cancer, the theory being that these cells undergo an epithelial to mesenchymal transition (EMT) in order to invade and metastasise (Falcioni, Antonini et al. 1997; Garzino-Demo, Carrozzo et al. 1998; Jin and Varner 2004; Gilcrease 2007). Integrins have also been implicated in angiogenesis with a potential role in the hypoxic environment of a proliferating tumour (Kim, Bell et al. 2000). Early reports of integrin expression in paediatric sarcomas of the small round blue cell type as well as some adult subtypes, indicated a heterogeneous expression of a variety of alpha and beta-integrins (Barth, Moller et al. 1995; Benassi, Ragazzini et al. 1998).

The findings in the present and previous studies provide the impetus to consider the over-expression of integrins alpha-2 and alpha-6 in STS in the context of potential therapeutic applications. Small molecule inhibitors of other integrins (ανβ3, ανβ5, α5β1)

~~156~~

~~Chapter 3: Candidate Gene Validation~~

~~are already in clinical trial phase, with others in preclinical development (Jin and Varner 2004).~~

~~3.4.1.3 NDRG1~~

The transcript of this candidate gene was shown to be over-expressed in the metastatic cell line GCT, but more significantly so in the high grade sarcoma cell line HT1080. This validated the findings of the gene expression arrays.

There is evidence that NDRG1 is induced by a variety of factors, including nickel and cobalt, DNA damage and hypoxia (Park, Adams et al. 2000; Salnikow, Costa et al. 2000). In other words, it may be modulated by cellular stress. NDRG1 expression has also been noted in other malignancies, with conflicting theories as to its role. It has been postulated to play a role both in cellular differentiation in colon cancer, as well as tumour progression (Guan, Ford et al. 2000; Wang, Wang et al. 2004). It has been implicated in malignant transformation in oral squamous cell carcinoma but in metastasis suppression in breast and prostate cancer (Bandyopadhyay, Pai et al. 2004; Chang, Wang et al. 2005; Bandyopadhyay, Wang et al. 2006). Its role, therefore, may be tissue-specific.

The expression pattern of NDRG1 manifest in this study may suggest a role in tumour progression in soft tissue sarcoma, if not metastasis. Certainly, its upregulation in hypoxic conditions may explain its increased expression in rapidly proliferating tumours. To date, NDRG1 expression has not been studied in STS. The findings in this thesis thus represent a novel finding in the context of tumour progression in STS.

~~3.4.1.4 MAP4K4 and other MAP kinases~~

Differential expression of MAP4K4 was of borderline statistical significance on qRT- PCR. Other MAP kinases (MAPK13 and MAP3K7) were also overexpressed in the high grade and metastatic cell lines on gene expression arrays. This signal transduction pathway may be utilised by a variety of growth factors and molecules, including EGFR and integrins in malignant transformation and tumour progression in STS.

~~157~~

~~Chapter 3: Candidate Gene Validation~~

~~The Erk/MAPK pathway will be examined further in the context of EGFR downstream signalling in Chapter 4, using the tissue microarray technique.~~

~~3.4.1.5 ACVR2B~~

Activin receptor 2B was most highly expressed in the metastatic cell line GCT on qRT- PCR. As noted in the introductory section (Section 3.1.5.1), dysregulation of activin expression has been associated with prostate and breast malignancy. Interestingly, recent gene profiling studies on soft tissue sarcomas have reported ACVR1B to be differentially expressed between primary and metastatic leiomyosarcomas (Lee, John et al. 2004), and ACVR2A to be overexpressed in gastrointestinal stromal tumours (Segal, Pavlidis et al. 2003). It is possible therefore that the activin group of receptors, which are related to the TGFβ family of receptors, are involved in tumour progression in sarcoma.

~~3.4.1.6 Cyclin D1~~

Cyclin D1 was over-expressed in the high grade cell line HT1080 on in qRT-PCR compared to the low grade sarcoma cell line. Statistical analysis on the relative expression ratio of this transcript revealed an F ratio of 16.82 with a p value of 0.001.

A number of studies on soft tissue sarcomas (STS) have noted CCND1 over-expression in 21-59% of the primary tumours examined (Creager, Cohen et al. 2001; Kim, Cho et al. 2001; Fritz, Schubert et al. 2002; Yoo, Park et al. 2002; Horvai, Kramer et al. 2006; Saito, Oda et al. 2006). CCND1 expression in some cases, could not be found to correlate with tumour grade (Creager, Cohen et al. 2001; Yoo, Park et al. 2002). In this study, there was relative underexpression of CCND1 in the sarcoma cell lines compared to the normal fibroblast cell line MRC5.

The findings in the present study and the conflicting reports noted above, warrant further investigation into the role of CCND1 in mitogenic signalling within STS. However, due to time constraints, this was not investigated further in this thesis.

~~158~~

~~Chapter 3: Candidate Gene Validation~~

~~3.4.2 Concordance with Gene Expression Arrays~~

In assessing statistical significance, if p<0.05 is used, the integrins ITGA2 and ITGA6, NDRG1, EGFR, CCND1 and ACVR2B were differentially expressed, in concordance with the gene expression arrays. HMGB1 and PDGFRA were not differentially expressed, also in concordance with the microarray experiments. In other words, six of the ten genes selected validated the microarray findings, while a seventh gene, MAP4K4, showed borderline statistical significance (p = 0.06) and two other genes were found not to be differentially expressed in either the gene expression arrays or the qRT-PCR experiments (Figure 3.17). This is better than previous studies reporting a concordance rate of between 0.6 - 0.67 (Larkin, Frank et al. 2005).

~~Microarray and qRTPCR Concordance~~

~~2 6 R = 0.8738~~

~~2 2 Ct)~~

~~2 R = 0.812 0 -4-3-2-10123 -2~~

~~qRTPCR (log -4~~

-6

-8

~~-10~~

~~M (log2 red/green) Mean HT1080 Mean GCT Linear (Mean HT1080) Linear (Mean GCT)~~

Figure 3.11 Microarray and qRT-PCR Concordance The concordance for two of the sarcoma cell lines is presented graphically above for the genes validated using qRT-PCR. Each coloured spot represents the log2 transformed expression ratios on the gene expression arrays (y axis) versus that on quantitative RT-PCR in real time (x axis) for each gene. The concordance rate is given by the R2 values for the trendlines. In other words, for the above cell lines, there was > 80 % concordance rate between the gene expression arrays and qRT- PCR. There are a few potential reasons for non-concordance between microarray and qRT- PCR data for some of the genes:

~~159~~

~~Chapter 3: Candidate Gene Validation~~

• Small abundance in mRNA copy number. CCND1 and PDGFRA, for instance, were underexpressed in all three sarcoma cell lines, although to different degrees. The technique of microarray is useful for establishing medium to high gene expression differences. Measurement of gene expression differences at the extreme ends of high and low expression is less accurate. • Gene annotation and splice variants. Validation is dependant on accurate and complete gene annotation of the oligonucleotides printed on the array. Annotation, in some cases, may be incomplete, with the existence of unannotated splice variants (Larkin, Frank et al. 2005; Sherlock 2005). It is possible, therefore, to have the oligonucleotide on the microarray targeting one splice variant and the RT-PCR primer targeting another, leading to discordant results. Both EGFR and ITGA6 have splice variants. At the time of designing the primers for qRT-PCR, the splice variants for ITGA6 were not known. • The dynamic range of the oligonucleotide arrays and qRT-PCR platforms are different. The fluorescence intensity for qRT-PCR is greater than microarray. Thus the degree of differential expression may vary between the two platforms. This might explain some of the differences in signal intensity between microarray and qRT-PCR for CCND1 and PDRGFA. • Primers for qRT-PCR are designed based on whole transcript whereas the oligonucleotide sequence printed on the array is "randomly" designed by algorithm. If the RT-PCR primer design had been based on the oligonucleotide that was printed on microarray, the primers would have detected the exact transcript as the microarray experiment. However, the argument can be made that designing the primers based on the whole transcript of the candidate gene ensures that the actual signal detected is not random (false discovery). In the case of TGFB1, only partial cds were printed on the oligonucleotide array (M38449) whereas the qRT-PCR primer design was based on the whole transcript (NM_000660).

~~160~~

~~Chapter 3: Candidate Gene Validation~~

~~3.5 CHAPTER SUMMARY~~

• Real time RT-PCR yielded > 70 % concordance with the gene expression arrays for the candidate genes selected, and this is in keeping with published rates of concordance. Gene expression profiling using cell lines is therefore a valid technique for hypothesis generation in the study of soft tissue sarcomas, providing a powerful tool for candidate gene selection. • This study confirmed differential expression of EGFR, indicating that it may have diagnostic, prognostic and therapeutic significance. The potential therapeutic implications can be appreciated as small molecules targeting EGFR and integrins are already in clinical trial phase for the treatment of certain malignancies. • The differential expression of other biomarkers such as NDRG1, ACVR2B, CCND1 and the integrins ITGA2 and ITGA6 were also validated. • The diagnostic and prognostic significance of EGFR was investigated further in Chapter 4, by creating a tissue microarray of soft tissue sarcoma samples, examining protein expression in these tumours and carrying out clinical correlation.

~~161~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

~~2" !~~

~~4. EGFR and SIGNAL TRANSDUCTION in STS: CORRELATION with PATIENT OUTCOME~~

~~162~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

~~4.1 INTRODUCTION~~

For any given gene, levels of mRNA expression and protein expression may not correlate, due to factors such as translational regulation (rather than transcriptional) and post-translational modification of proteins. The exact frequency with which protein expression correlates with levels of gene expression, as measured on arrays, is not known, but is estimated by users of these technologies to be in the region of 50% (Chuaqui, Bonner et al. 2002). The sources of variation in gene expression profiling using oligonucleotide microarrays were discussed in Chapter 2 and the need for validation studies presented (Section 2.4.6). The strength of gene expression arrays lies in their ability to examine global patterns of gene dysregulation and aid in generation of hypotheses. However, it is essential to use clinical samples, in order to determine the broader application of the gene expression data in clinical practice.

This chapter examines the protein expression of EGFR in a tissue microarray of 92 formalin-fixed soft tissue sarcoma specimens of varying histologic grade from 89 patients. Activated EGFR expression was also examined, in order to determine whether the receptor is activated in these tumours. Three downstream signal transduction pathways that may be activated by EGFR are also assessed. EGFR was chosen for further study on the basis of the following: • EGFR gene expression increased across the sarcoma cell lines of increasing tumour grade on gene profiling in Chapter 2. A number of its ligands were also overexpressed in the higher grade cell lines. • The above finding was further validated on real time RT-PCR in Chapter 3. • EGFR has not been studied in the context of tumour progression and metastasis in soft tissue sarcoma. To our knowledge, the expression of the activated form of the receptor has not been examined in these tumours, nor have the possible downstream mechanisms at play. • EGFR clearly represents an attractive therapeutic target as several inhibitors have already been developed.

~~163~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

~~4.1.1 Advantages of tissue arrays~~

Tissue microarray (TMA), as first described (Kononen, Bubendorf et al. 1998), enabled the precise “arraying” of core biopsies taken from formalin-fixed paraffin- embedded “donor” blocks onto a new “recipient” block. In that study, 0.6 mm cores were taken from breast cancer specimens and in situ hybridisation (ISH) carried out for three known oncogenes. The advantages of this new technique, as noted subsequently in a number of reviews (Hoos and Cordon-Cardo 2001; Rimm, Camp et al. 2001; Skacel, Skilton et al. 2002; Packeisen, Korsching et al. 2003; Nilbert and Engellau 2004) are discussed below.

~~4.1.1.1 High Throughput~~

High throughput techniques have provided a major advance for immunostaining. Hundreds of 5 μm sections of the same set of samples can be obtained from one master block. This enables the screening of molecular targets on a large number of archival clinical samples. It also contrasts with other high throughput technologies, such as cDNA or oligonucleotide arrays, where fresh tissue is required for optimal results (Kallioniemi, Wagner et al. 2001).

~~4.1.1.2 Preservation of Archival Tissue~~

The use of 0.6-1mm cores means that there is minimal loss of tissue from each donor block. There is no minimal of integrity, and the donor block can be sectioned for further studies, or used for constructing more tissue arrays.

~~4.1.1.3 Cost Effectiveness and Reproducibility~~

With multiple tumours arrayed on a single slide, a fraction of the reagents that would be required for conventional staining of full sections are utilised, rendering TMA a cost effective research tool. By performing immunostaining on multiple samples on a single slide, intralaboratory variability can be reduced (Packeisen, Korsching et al. 2003).

~~4.1.2 Validity of TMA~~

Two concerns have been raised with regard to the validity of the technique: that the 0.6 mm cores taken may not be representative of the entire tumour and that archival specimens that may have lost their antigenicity would be used. A number of validation

~~164~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

studies have addressed these issues (Table 4.1). These studies aimed to determine the optimal number of cores (Camp, Charette et al. 2000; Gillett, Springall et al. 2000; Torhorst, Bucher et al. 2001; Rubin, Dunn et al. 2002; Skacel, Skilton et al. 2002), the optimal size of cores (Skacel, Skilton et al. 2002), as well as whether cores were able to represent the heterogeneous nature of tumours (Engellau, Akerman et al. 2001; Torhorst, Bucher et al. 2001; Hsu, Nielsen et al. 2002; Jourdan, Sebbagh et al. 2003). Antigenicity of archival specimens, inter- and intralaboratory reproducibility have also been proven (Camp, Charette et al. 2000; Hsu, Nielsen et al. 2002; von Wasielewski, Mengel et al. 2002; Dolled-Filhart, Camp et al. 2003). These are presented in greater detail in Appendix IV.

~~4.1.3 Application of Tissue Microarray technology~~

~~4.1.3.1 Tissue Microarray in Epithelial Malignancy~~

Table 4.2 summarises selected studies carried out on epithelial tumours using TMA and are described in greater detail in Appendix IV. The greatest utility is achieved when degree of immunostaining or expression is able to be correlated with clinical outcome.

~~4.1.3.2 Tissue Microarray in Soft tissue sarcoma~~

TMA studies involving STS are summarised in Table 4.3 and presented in greater detail in Appendix IV. TMA studies on synovial sarcoma (SS) and gastrointestinal stromal tumours (GISTs) cited earlier gene profiling work, identifying potential biomarkers predictive of histologic subtype (Nielsen, Hsu et al. 2003; West, Corless et al. 2004). Elsewhere, markers including Ki-67, beta-catenin, and CD44 were empirically chosen based on previous literature (Engellau, Bendahl et al. 2005). The three previous studies examining EGFR expression in sarcoma concentrated either on synovial sarcomas (Allander, Illei et al. 2002; Nielsen, Hsu et al. 2003) or rhabdomyosarcomas (Ganti, Skapek et al. 2006). None of these investigated EGFR activation or signal transduction across the histologic subtypes.

Aberrant Wnt-β-catenin signaling and overexpression of Cyclin D1 in SS has also been examined by TMA (Saito, Oda et al. 2000; Saito, Oda et al. 2001; Ng, Gown et al. 2004). Horvai et al evaluated the correlation between β-catenin and Cyclin D1

~~165~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

expression on an array incorporating 82 SS but no significant difference was noted between primary and metastatic tumours (Horvai, Kramer et al. 2006). Another member of the Wnt signaling pathway, Transducin Enhancer of Split 1 (TLE1) was overexpressed in SS (Terry, Saito et al. 2007), confirming findings in previous studies (Ng, Gown et al. 2004; Pretto, Barco et al. 2006).

~~166~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

Table 4.1 Tissue Microarray Validation Studies in Other Malignancies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

~~167~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

Table 4.2 Tissue Microarray studies in Other Malignancies 832&-0D-30, *?03'113#'-+ 0)#01: *'" 2'-,-$! <8 6*'%-V 00 71 7-,-,#,D < 2#"_[Z 0# 12* ,!#0 ?*Q%-(TQ**< SQ ,"-2 60'%', *. .#0 0*3,"D<*_RR0# 12* ,!#0X7Q%-(T! <8 00 71S!#***',#1 %#"#,$ *) <-D_RS 0# 12* ,!#0 **< SQ-7S ! <8 00 71SSV(*8ST! 00'#01 -,,'6 .<8_RS 0# 12* ,!#0 SY/TU +.*'$'! 2'-, *&% 00 71S!#***',#1 )0#21-4<**',* (#1_RV0# 12* ,!#0US+ 0)#01-+.'0'! *!&-'!#-$ ,2'-"'#1 3 #,"-0$/D<*_[[.0-12 2#* ,!#0&".TQ%.TY! <8 00 71S*F(TT(6#,-%0 $21 -311#1 * (#1_RT .0-12 2#* ,!#0 SRR. ! <8 00 71S*F(TT(6#,-%0 $21 & , 1#) 0 ,QTT< 230#_RS.0-12 2#* ,!#0%#.1',Q.'+S! <8 00 71S!#***',#123+-301 -!&%8D._[[(#, *!#*** ,!#0:'+#,2',! <8 00 71S!#***',#1 **',#,/* (#1_RR&*'-+ 1&".T! <8 00 71S.0'+ 070#!300#,223+-30 ,!V* 0 7- * (#1_RT * ""#0* ,!#0 -V! "�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

~~Chapter 4: EGFR and Signal Transduction in STS~~

Table 4.3 Tissue Microarray studies in Sarcoma 832&-0D-30, *?03'113#'-+ 0)#01: *'" 2'-,-$! <8 6*'%-V 00 71 6.0#11'-,', 0!-+ 8** ,"#0:8D._RT7,-4' * 0!-+ -(TQ&".T! <8 00 71SS[23+-301 <'#*1-,368D._RU7,-4' * 0!-+ -&"(Q8//T! <8 00 7.3*'1&#"1#. 0 2#*73 & ,2'( -". 2&_RX -(8( -&"(Q-(T 62� 32&-01_123"'#1-,(!#***',#1 6.0#11'-,-$ 2�&'-+ 0)#01', 0!-+ F#12(8D._RV3 6&S',&31! <8 00 7.3*'1&#"1#. 0 2#*73 -,%#** 3D%3+. 2&-*_RW37'XYQ.V%*7!-.0-2#',Q* VVQβV! 2#,', <-%#,##6.0#11'-, 00 71 %-04 '8- 8./_RX 7,-4' * 0!-+ **< SQβV! 2#,', <-%#,##6.0#11'-, 00 71 &8_RX3( . 2&5 7.0-2#',1#<-%#,##6.0#11'-, 00 71 3#007D 8D._RY7,-4' * 0!-+ 3/-S62� 32&-01_123"'#131#" 3S-$22'113#1 0!-+ R 6&SS%7.-2'! *.0-2#',"/DSRTXSQ2#0+#"b '1!-4#0#"6,&3Sc 7 32&-01R&3S& 120-',2#12', *120-+ *23+-30R#( . 2&5 7 .0-2#',1S**< SQ* 7VQ* 7XQ-T"SQ.SX ,".TYR-(8(S-+ 07-, *8*4#-* 00& "-+7-1 0!-+ . #"' 20'!3R3<'#*1-,36#2 *Q/ ,!#2TRRT

~~169~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

~~4.1.4 The Present Study~~

In this thesis, as shown in Chapter 2, gene expression arrays identified a large number of transcripts to be over-expressed in the high grade and metastatic cell lines compared to the low grade cell line. Some of these were validated using qRT-PCR in Chapter 3. EGFR was found to have increased expression in the high grade cell line and even greater expression was seen in the metastatic cell line. Based on this finding, EGFR expression and some of its downstream signalling molecules are therefore evaluated on a tissue microarray of STS specimens of varying grade and metastatic potential.

~~4.1.4.1 EGFR and Signal Transduction~~

The human epidermal growth factor receptor is a 170 kDa receptor tyrosine kinase (RTK), the activation of which mediates cell proliferation, survival, adhesion, migration, angiogenesis and differentiation (Ciardiello and Tortora 2001; Yarden 2001; Jorissen, Walker et al. 2003; Mendelsohn and Baselga 2003). Ligand binding by growth factors such as EGFR, TGFα, amphiregulin and epiregulin triggers autodimerisation and activation of the receptor tyrosine kinase. Downstream signal transduction of EGFR may involve a number of cascades, including the mitogen-activated protein kinase (MAPK)24, phoshatidylinositol 3-kinase (PI3K) and the Janus tyrosine kinase (JAK)/signal transducers and activators of transcription (STAT) pathways (Vivanco and Sawyers 2002; Jorissen, Walker et al. 2003; Bianco, Melisi et al. 2006).

In this study therefore, EGFR and activated EGFR were examined, together with a phosphorylated (activated) downstream molecule from each of the above cascades, p44/42MAPK, pAkt and pSTAT3, in order to determine whether • EGFR expression was increased in STS

24 The MAPK cascade was discussed in Chapter 2 (Section2.4.2.5). The overexpression of MAP4K4 in the high grade and metastatic sarcoma cell lines was confirmed by real time RT-PCR in Chapter 3 (Section 3.4.1.4) 170

~~Chapter 4: EGFR and Signal Transduction in STS~~

• EGFR expression correlated with the histologic grade of the tumour • EGFR expression correlated with prognosis • EGFR expression correlated with increased expression of one or more of the downstream signal transduction molecules, thus indicating a mechanism of action in these tumours.

~~4.1.4.2 Erk/MAPK Signalling~~

Activation of Grb2, which is bound to Sos, triggers the ERK/MAPK pathway (Jorissen, Walker et al. 2003). Ras activation in turn results in Raf-1 activation, leading to phosphorylation of the MAP kinases. Also known as MAPK3 and MAPK1, p44 and p42 MAPK (Erk1 and Erk2) are intermediaries in the signal transduction cascade (Hunter 1995; Marshall 1995). Expression of the activated form of Erk1 and Erk2 (p44/42MAPK) is investigated using the TMA of STS specimens.

~~4.1.4.3 PI3K-Akt Signalling~~

Akt, also known as protein kinase B (PKB) or Rac, is a serine/threonine protein kinase thought to mediate cell growth and proliferation through the mammalian target of rapamycin (mTOR), either directly, or by inactivating its inhibitors (Bianco, Melisi et al. 2006). Other apoptotic factors such as Bad and caspase 9 are inactivated, promoting survival (Cardone, Roy et al. 1998). PTEN is known to inactivate Akt. This signalling cascade is frequently dysregulated in various cancers, with aberrations including: constitutive activation of Akt or mTOR; inactivation of PTEN; or, induced activation of Akt by growth factor receptors such as EGFR. The phosphorylated form of Akt (pAkt) is examined by immunostaining in this chapter.

~~4.1.4.4 STAT3 Signalling~~

The Signal Transducers and Activators of Transcription (STAT) proteins are cytoplasmic proteins activated by tyrosine phosphorylation either through cytokine receptor associated kinases or growth factors such as EGFR (Garcia and Jove 1998; Han, Hwang et al. 2005). For STAT3, phorphorylation of the tyrosine residue Tyr705 is required for activation (Calo, Migliavacca et al. 2003). Following phosphorylation and dimerisation, cytoplasmic to nuclear translocation occurs. DNA binding ensues, triggering the activation of a variety of transcription factors (Darnell, Kerr et al. 1994;

~~171~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

Ihle 1995; Wen, Zhong et al. 1995). Direct STAT3 activation by EGFR has also been reported (Barnes and Kumar 2003). It is postulated that STAT3 may itself function in some cases as an oncogene by dysregulation of the signalling cascade but in others as a tumour suppressor (Bowman, Garcia et al. 2000; Bromberg and Darnell 2000). STAT3 has been shown to be constitutively activated in some epithelial tumours (Garcia and Jove 1998; Catlett-Falcone, Landowski et al. 1999).

~~NRG Extracellular EGF EREG 8(-& BTC TGF~~

~~EGFR P13 JAK Connexin~~

~~SOS Grb2 mTOR AKT-DK1 Myc RAS Cyclins STAT3 STAT3 GTPase CDKs Cyclins~~

~~Raf Transcription~~

~~Transcription Factors ERK/ MAPK Factors~~

~~Nucleus Cytoplasm~~

Figure 4.1 EGFR signalling pathways Some of the ligands that can bind to and activate the receptor, including EGF, TGFααα, epiregulin (EREG), neuregulin (NRG), amphiregulin (AREG) and betacellulin (BTC) are shown. Ligand independent receptor activation is known to occur in some tumours. EGFR autodimerises and triggers a signalling cascade. Ras/Raf/MAPK , PI3K/Akt and JAK-STAT (shown above) are the pathways known to be involved in EGFR signalling. These are investigated in this chapter. CDK: Cyclin dependant kinase; Grb2: growth factor receptor bound protein 2; JAK: Janus kinase; mTOR: Mammalian target of rapamycin; Raf: mutant retroviral transforming agent; SOS: son of sevenless; STAT: Signal transducer and Activator of Transcription.

~~172~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

EGFR over-expression alone, however, does not necessarily predict activation of downstream signalling pathways. Different mutations within the receptor may determine which pathways are activated, or conversely, there may be mutations downstream of EGFR. For these reasons, the expression of EGFR and activated EGFR, as well as that of the activated downstream mediators pAkt, p44/42 MAPK and pSTAT3 was examined by immunohistochemistry in this chapter.

~~4.2 METHODS~~

~~4.2.1 Selection of archival specimens~~

All donor specimens were reviewed and graded independently by an anatomical pathologist. 93 soft tissue sarcoma specimens resected at Prince of Wales Hospital, Sydney, between January 1999 and March 2005 were selected. This included all specimens with adequate archival paraffin-embedded tissue available for sampling. For the purposes of constructing this tissue array, carcinosarcomas, gastrointestinal stromal tumours (GISTs)25 and tumours with “controversial” histologic diagnoses were excluded. Areas representative of the overall tumour grade were identified by our pathologist and selected and marked on the haematoxylin and eosin (H&E) stained slides corresponding to the donor blocks.

~~4.2.2 Clinic Database~~

A prospective database has been maintained at the divisions of surgical oncology and radiotherapy in Prince of Wales Hospital. All clinical information and follow up data has been recorded in this database (Appendix IV).

25 As discussed in Chapter 1 Section 1.4.2, the French histologic grading system (FNCLCC) does not apply to GISTs, which are described as being malignant on the basis of size, mitotic rate and metastases. 173

~~Chapter 4: EGFR and Signal Transduction in STS~~

~~4.2.3 Construction of TMA master blocks~~

The array was constructed using the ATA100™ Advanced Tissue Arrayer (CHEMICON International, Inc., Temecula, CA). 1 mm cores in triplicate for each tumour were taken. Additional cores were taken from randomly selected tumours as internal controls for intratumour variability in each of the broad histologic subtypes. Details of the construction are provided in Appendix IV. Four master blocks were initially created. Block A comprised leiomyosarcomas (LMS), Block B, liposarcomas (LPS), Block C, malignant fibrous histiocytomas (MFH) or pleomorphic sarcomas (PMS) and on Block D, the less common STS such as malignant peripheral nerve sheath tumours (MPNST), synovial sarcomas (SS), angiosarcomas (AS) and epithelioid sarcoma (EMS).

~~4.2.4 Assessment of integrity of tissue cores on TMA master blocks~~

Sections were cut from these master blocks and H&E stained for histologic verification. The donor blocks for any “missing” cores were re-examined and two further blocks E-F constructed, obtaining extra cores from specimens arrayed on A-B and C-D.

~~4.2.5 Immunohistochemistry~~

The immunostaining techniques used after a series of optimisation experiments are summarised in Table 4.4. During the optimisation process, normal prostate and breast cancer tissue were used as positive controls for EGFR (Clone 31G7, Zymed Laboratories, Inc., CA, USA), while colon cancer specimens were used for activated EGFR (MAB3052, Chemicon International, Inc., Temecula, CA). Colon cancer was also used as the positive control for p42/44 MAPK and pSTAT3, while prostate cancer sections were used as positive controls for pAkt (Cell Signaling Technology, Inc., Danvers, MA, USA). Subsequently, appropriate sarcoma specimens were used as positive controls. Phosphate buffered saline (PBS) and matched irrelevant mouse or rabbit IgG1 antibodies were used as the negative controls. The antigen retrieval method for EGFR was pepsin digestion. Activated EGFR, pAkt and p44/42MAPK were microwaved in citrate buffer and pSTAT3 was microwaved in EDTA as per manufacturer’s instructions. The incubation period was standardised for all antibodies to overnight at 4°C. The detailed protocol is given in Appendix IV.

~~174~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

~~4.2.6 Scoring of Immunostaining~~

Scoring was carried out independently by two researchers (Romi Das Gupta and Jia Lin Yang) who were blinded to the clinicopathologic characteristics of the patients. Where there were discrepancies in the scoring, the cores were reviewed together and a consensus score was assigned. The average score was taken for replicates. A core was considered to be uninterpretable if < 10% of it remained intact on the slide. Where there were only two cores with discrepant but positive staining (i.e. 1+ or 2+), the higher score of the two was taken. Where only 2 cores were available for scoring and was positive in one core and negative in the other, the higher score was recorded for the monoclonal antibodies and the lower score for the polyclonal antibody. Semi- quantitative criteria were used for scoring the slides and were as follows: 0 (Negative): No positive staining. 1 (Weak positive): Intense staining in < 10% or diffuse staining in < 50% of core 2 (Strong positive): Intense staining in 110% or diffuse staining in 150% or core The above scoring is similar to semi- quantitative methods described in other tissue microarray studies where EGFR staining or staining for the downstream phosphorylated signal transduction antibodies were carried out (Dolled-Filhart, Camp et al. 2003; Nielsen, Hsu et al. 2003).

~~4.2.7 Clinicopathologic correlation and Survival analysis~~

The Kruskal Wallis test or Mann Whitney U test and chi-square tests were applied to assess correlation between biomarker expression and histologic grade and subtype. The chi-square test was also used to assess correlation between EGFR, activated EGFR and the downstream signal transducers.

Cox (proportional hazards) regression model was employed to determine independent prognostic factors. This systematically tests for different combinations of covariates. The end points for survival analysis were disease free survival (DFS), overall survival (OS) and death from disease. The time to these end points was calculated from the day of commencement of treatment. As there were only 9 local recurrences (LR), this group was too small to statistically analyse separately. Calculation of DFS and OS (cancer specific survival was similar to OS and was not shown since only one patient died of other course) was carried out using the Kaplan Meier method with loss of follow up

~~175~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

censored. Within subgroups, variables were compared by Log-rank test. This method tests the null hypothesis that there is no difference between the subgroups in the probability of an event (death, in the case of a survival analysis). The assumption made, is that censoring26 is unrelated to prognosis. Statistical values or p < 0.05 were considered significant. The analysis was carried out using the SPSS/Win (SPSS, Inc, Chicago, IL, USA) statistical software package.

~~26 Censorship, in this context, refers to loss of follow up of patients, or patients being alive at the end of the study period, where the outcome or event being measured is death. 176~~

Chapter 4: EGFR and Signal Transduction in STS Table 4.4 Antibodies Selected for Immunohistochemistry on TMA ",2' -"7 -30!# )'*32'-, ",2'%#,#20'#4 * ,!3 2'-, T°°°",2' -"7 +'13 *'1 2'-, -&"( W7+#" SSTR .#.1',"'%#12'-, 64#0,'%&2V°* ,2'V+-31#8*:#!2 12 ', 8 8!2'4 2#"-&"( *&#+'!-, SSTR SH*'20 2# 3$$#0 . 64#0,'%&2V°* ,2'V+-31#8*:#!2 12 ', 8 .8)2(. *#**'%, *',% SSTR SH*'20 2# 3$$#0 33 64#0,'%&2V°* ,2'V0 '2 8*:#!2 12 ', 8 .VV VT8.7(. *#**'%, *',% SSWR SH*'20 2# 3$$#0 . 64#0,'%&2V°* ,2'V0 '2 8*:#!2 12 ', 8 -0)S -0)T .383U( *#**'%, *',% SSTR S+- 38.%Z 64#0,'%&2V°* ,2'V0 '2 8*:#!2 12 ', 8 S-31#+-,-!*-, * ,2' -"7Q(.S( '2.-*7!*-, * ,2' -"7Q(S( '2+-,-!*-, *Q.S.&-1.& 2# 3$$#0#"1 *',#Q 33SSH30'1 3$$#0#"1 *',# RTS$35##,VTRQ- 38S-2&7*#,#"' +',#2#20 !#2 2#"'1-"'3+"#&7"0 2#Q 8SUQUV"' +',- #,8'"',#T#!-," 07 ,2' -"'#1Q 8*:#!2 12 ', ," 85#0#- 2 ',#"$0-+:#!2-0/ -0 2-0'#1Q30*',% +#Q*8T

~~177~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

~~4.3 RESULTS~~

~~4.3.1 Tumour Histology and Grade~~

Of the 92 STS used for the TMA, 17 were leiomyosarcomas (LMS), 22 were liposarcomas (LPS), 30 were pleomorphic sarcomas (PMS) or malignant fibrous histiocytomas (MFH) and 23 tumours fell into the category of less common STS such as synovial sarcomas (SS), malignant peripheral nerve sheath tumours (MPNST), and rhabdomyosarcomas (RMS). This last category also included one patient deemed to have a peripheral neuroectodermal tumour (PNET) and another with an embryonal RMA (ERMS), which are usually more common in the paediatric population.

25 tumours were defined as low grade, 8 were of intermediate grade, 46 were high grade and 11 were metastases. 13 tumours had been subjected to neoadjuvant treatment, 9 of which still had considerable viable residual tumour at the time of resection judged to be of high grade. In 2 cases the grade was difficult to ascertain on the basis of radiotherapy effects. There were 9 local recurrences (LR), 3 of these in low grade tumours one in an intermediate grade and the remainder, in high grade tumours.

~~4.3.2 Integrity of tissue cores~~

There were a total of 298 cores from 92 tumours arrayed on blocks A to D. These were sectioned and the haematoxylin and eosin (H&E) stained slides were then compared to the template. 54 cores were found to be “missing”, indicating that 81.9% of the cores sampled were successfully arrayed. The remaining “missing” samples were re-arrayed on blocks E and F. The average number of cores per specimen for each antibody is given in Table 4.5. TMA Master Block B, comprising LPS, proved the most difficult to section. Hence, for this block, two sections were placed on each slide for the activated EGFR, pAkt and p44/42MAPK antibodies. There were no specimens that needed to be excluded from analysis for activated EGFR and pAkt on the basis of there being no cores available to interpret. EGFR and p44/42MAPK each had one tumour excluded from analysis, while pSTAT3 had three specimens excluded. Only 3% of specimens had a single core available for analysis. This compares favourably to published data (see Appendix IV for detail).

~~178~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

Table 4.5 Integrity of Tissue Cores Available for Analysis 8,2' -"7 *-0#1.#0.#!'+#, S*-0#84 '* *# <-*-0#184 '* *# -&"(VTZSTS 8!2'4 2#"-&"(VTTYUR .8)2VTTZSR .VV VT8.7VTTUTS .383UUTVZXU 3&#$'012!-*3+,',"'! 2#12&# 4#0 %#,3+ #0-$!-0#1 4 '* *#$-0 , *71'1$-02&#[T1.#!'+#,1 31#" 2- !-,1203!2 2&# 2'113# 00 71T 30'.*'! 2# !-0#1 5#0# ','2' **7 00 7#" 2&#, 1#!2'-,#" ," 12 ',#"5'2&%-2-#6 +',#2&#',2#%0'27-$2&#!-0#1T.#!'+#,1$-3,"2-& 4#!-0#1b+'11',%c 5#0# 0#V1 +.*#" ," 00 7#" -, 1#. 0 2# 0#!'.'#,2 + 12#0 *-!)Q 5&'*# 1-+# 1.#!'+#,1 5#0# 00 7#"$0-+"'$$#0#,2"-,-0 *-!)1-,2-2 +#0#!'.'#,2 *-!) 1',2#0, *!-,20-*1T%#,!#2&# 4#0 %# ,3+ #0-$!-0#1.#01.#!'+#, #6!##"1U$-0 **2&# ,2' -"'#1123"'#"T, 2�#+ ',',% !-*3+,1Q2&#,3+ #0-$1.#!'+#,1$-05&'!&-,*7-,#!-0#-0,-!-0#15#0# 4 '* *#$-0 , *71'1 0#0#!-0"#"T6,#"-,-01.#!'+#,Q !-0# '-.175'2&*'22*# 4 '* *#23+-302'113#!-3*",-2 #0#V 1 +.*#" ,"&#,!#,-!-0#15#0# 4 '* *#$-0U-$2&# ,2' -"'#1-&"(Q.VV VT8.7 ," .383UT

~~4.3.3 Intra-Tumour Variability/Heterogeneity~~

In each of the TMA master blocks, one or more tumours were sampled from different donor blocks, that is, different regions of the tumour. For the given patient or tumour, therefore, a minimum of six cores were arrayed onto the master block. Immunostaining scores were recorded, in order to determine whether there was variability in the staining pattern depending on the region of the tumour the cores were taken from. For TMA master block B, comprising liposarcomas, two tumours from two different patients were thus arrayed, providing two “internal controls” for this block. Two internal controls were also present in TMA Block C, where multiple samples from different regions of the tumour were sampled from a locally recurrent malignancy and a metastasis presenting in the same patient asynchronously. In TMA master block D, the tumour used as the internal control was sampled from the initial wide local excision, as well as from the re-excision of the margins performed some days later. The results of this analysis are summarised in Table 4.6. As can be seen, the staining patterns were largely concordant between the cores from different regions of the tumour.

~~179~~

Chapter 4: EGFR and Signal Transduction in STS Table 4.6 Internal Controls Assessing Intra-Tumour Heterogeneity 38 12#0*-!). 2'#,2 -,-0*-!)1++3,-12 ',',%*-,!300#,!#?#1 <- -&"(8!2'4 2#"-&"(.8.7.8)2.383U 8/ STQU <S&RT& ? ? ? ? /. TWQS[ ? ? ? ? ? U VQ X??? ? ? *. "% VQ/(TQU ? <T&RS& ? ? ? VQ#2[?>#? ? ? 62 W(Q3 ? <T&RS& ? ? <RRS& 38 12#0*-!)18V !-+.0'1#"3-$"'$$#0#,2&'12-*-%'!13 27.#1Q 1"'1!311#"',"#2 '*',2&##2&-"1#!2'-,VTTTUTUT/S/#'-+7-1 0!-+ 1R/.S /'.-1 0!-+ 1R . "%S .*#-+-0.&'! 0!-+ 1 *'%, ,2 "' 0-31 %'12'-!72-+ 1R 62S /#11 !-++-, 1 0!-+ 1 13!& 1 7,-4' * 0!-+ 1 Q 8,%'-1 0!-+ 1Q *'%, ,2.#0'.� *<#04#&# 2&33+-301.<31 ,"(& "-+7-1 0!-+ 1(T3&#. 2'#,21 0#"#V'"#,2'$'#" ,",3+ #0#"$0-+SVW $-0 2&# .30.-1# -$ 2&'1 3 *#T 3&# ,3+ #01 ', 0 !)#21 0#$#0 2- 2&# "-,-0 *-!)1 ," 2�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�#2 ',',%. 22#0,"'",-2!-,!30 #25##,2&# *-!)11 +.*#"$0-+2 +#23+-30Q2&#"'$$#0#,21!-0#1 0#%'4#,', 0 !)#21T

~~180~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

~~4.3.4 EGFR and Activated EGFR~~

~~4.3.4.1 EGFR~~

Immunostaining for EGFR was carried out using a mouse monoclonal antibody which recognises the extracellular domain of the EGF receptor. Immunopositivity was manifested as membranous staining. Intense or strong staining resulted in membranous as well as cytoplasmic staining. Conceptually, this is possible as activation of the EGF receptor, results in dimerisation and internalisation of the complex. Examples of the staining patterns are shown in Figures 4.3, 4.7 and 4.8.

33.7% demonstrated 1+ immunopositivity and 44.9% were scored as 2+. 60% of metastatic tumours and 58.7% of high grade tumours demonstrated 2+ immunopositivity, while 30% of metastatic tumours and 28.3% of high grade tumours were scored as 1+. Only 21.3% of tumours did not express EGFR. The number of specimens scoring 0, 1+ and 2+ in each histologic grade is shown below and the breakdown for positive and negative staining for EGFR is shown in Appendix IV.

Table 4.7 EGFR Immunoreactivity of All STS samples across Histologic Grade -&"( R S T 3-2 * %'12-*-%'! *-5%0 "# SS SR V TW &0 "# ',2#0+#"' 2#%0 "# S V U Z &'%&%0 "# X SU TY VX +#2 12 2'!%0 "#SUXSR 3-2 * S[ UR VR Z[

Raw data correlating EGFR expression with that of the signal transducers is presented below, while the statistical correlation of EGFR to other downstream signal transducers and histologic grade are presented in Sections 4.3.5 to 4.3.8. The relevant data is presented in Table 4.12 (p44/42MAPK), Table 4.15 (pAkt), Table 4.18 (pSTAT3) and Tables 4.20 – 4.21 (grade) and 4.22 (histologic subtype).

~~181~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

Table 4.8 Correlation of EGFR expression with Signal Transducers for All Samples 6.0#11'-, ++3,-12 ',',% !-0# RST, .VV VT#"'R X T S [ SWZVSY TZTTUXXX , S[ UT VS [T .")2 R Z S S SR STSSTSW TZTRUZXY , SZ UT VS [T . "URSXTSTZXW SUZSRTS TRTTV , S[USVR[R

~~4.3.4.2 Activated EGFR~~

Expression of the phosphorylated (activated) EGF receptor was detected using a mouse monoclonal antibody. This recognises the intracellular domain of EGFR, resulting in cytoplasmic staining. Intensity and distribution of cytoplasmic staining varied across the specimens, however nuclear or membranous staining was not seen. A selection of the cores stained for activated EGFR is shown in Figures 4.4 and 4.9.

There was a statistically significant correlation between EGFR and activated EGFR expression with a p value of 0.02 when the 3-point semi quantitative scoring systems were compared, and a p value of 0.05 when the groups were divided in binary fashion (positive/negative expression). This statistical significance was still maintained when patients who received neoadjuvant treatment were excluded from the analysis.

34.4% demonstrated 1+ immunopositivity and 51.1% were scored as 2+. 63.6% of metastatic tumours over-expressed activated EGFR with a score of 2+ and 36.4% scored 1+. 63% of high grade tumours were scored as 2+ and 40% of them scored 1+. Only 14.4% of tumours did not express activated EGFR. There was a statistically significant correlation between tumour grade and activated EGFR expression and this is presented in Tables 4.20 and 4.21.

~~182~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

Table 4.9 Activated EGFR expression in all STS samples across Histologic Grade 8!2'4 2#"-&"( R S T 3-2 * %'12-*-%'! *-5%0 "# ST X Y TW &0 "# ',2#0+#"' 2#%0 "# S V U Z &'%&%0 "# R SY T[ VX +#2 12 2'!%0 "# R V Y SS 3-2 * SU US VX [R

Activated EGFR expression was correlated to the expression of the activated downstream signal transduction molecules p44/42MAPK, pAkt and pSTAT3. The raw data is shown below while the statistical correlation is presented in Sections 4.3.5 to 4.3.7. The corresponding data is shown in Table 4.13 (p44/42MAPK), Table 4.16 (pAkt) and Table 4.19 (pSTAT3). The relevance of activated EGFR expression to clinical outcome in this group of patients with soft tissue sarcoma is considered in Section 4.3.10.

Table 4.10 Correlation of Activated EGFR expression with Signal Transducers in All Samples "!2'4 2#"6.0#11'-, ++3,-12 ',',% !-0# RST, .VV VT#"'R Z S R [ SV Y X SY TSTVVSXX , SU UT VY [T .")2 R Z S S SR S V W X SW TSTWVSXY , SU US VZ [T . "U R SS TT UT XW S T Y ST TS T R S U V , SU UR VY [R

~~183~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

~~4.3.5 p44/42 MAPK (Erk1/Erk2)~~

The rabbit polyclonal antibody used in this study detects activated levels of Erk1 and Erk2 protein in the MAPK cascade (Figure 4.1). Activation by phosphorylation of these signal transducers results in their cytoplasmic to nuclear translocation. Immunopositivity for the phosphorylated forms of Erk 1 and Erk 2 (p44/42MAPK) were therefore manifest as cytoplasmic and nuclear staining, as shown in Figures 4.4 and 4.7 to 4.9.

16.9% demonstrated 1+ immunopositivity and 73% were scored as 2+, as shown in the table below. 100% of intermediate grade and metastatic tumours and 97.9% of high grade tumours expressed p44/42MAPK. 10.1% of tumours did not express p44/42MAPK. The correlation with tumour grade was significant and is presented with all other biomarkers examined in Tables 4.20 and 4.21.

Table 4.11 p44/42 MAPK (Erk1 & Erk2) Immunoreactivity in all STS samples .VV VT8.7 R S T 3-2 * %'12-*-%'! *-5%0 "# Z SR Y TW &0 "# ',2#0+#"' 2#%0 "# R U W Z &'%&%0 "# S S VV VX +#2 12 2'!%0 "# R S [ SR 3-2 * [ SW XW Z[

The expression of p44/42MAPK was also correlated to that of EGFR and activated EGFR, as a means of assessing coregulation of these proteins (Raw Data: Tables 4.8 and 4.10). Table 4.12 shows the chi-square tests for statistical significance for EGFR and p44/42MAPK while Table 4.13 shows the same for activated EGFR and p44/42MAPK. As can be appreciated, p44/42MAPK expression correlates with the expression of both EGFR and activated EGFR. This also holds when the patients subjected to neoadjuvant treatment are excluded from the analysis.

~~184~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

Table 4.12 Correlation between EGFR and p44/42MAPK Expression (#- "(34 ,2 1#16!*3"#" "** 1#1 ," 817+.T '%T 817+.T '%T .VV VT#"' : *3# "$ TV1'"#" : *3# "$ TV1'"#" .# 01-,*&'V/3 0# SZTY[T SZTRUR V RTRRS V RTRRS /')#*'&--"( 2'- SXTRWW V RTRRU SXT[RX V RTRRT /',# 0V 7V/',# 0 SWTTSS S RTRRR SWTYSY S RTRRR 811-!' 2'-, <-$: *'"* 1#1 Y[ [T

Table 4.13 Correlation between Activated EGFR and p44/42MAPK Expression (#- "(34 ,2 1#16!*3"#" "** 1#1 "!2'4 2#" ," 817+.T '%T 817+.T '%T .VV VT#"' : *3# "$ TV1'"#" : *3# "$ TV1'"#" .# 01-,*&'V/3 0# WTTY[S VXTWRT V RTRRR V RTRRR /')#*'&--"( 2'- UZTUTV V RTRRR VTTZWR V RTRRR /',# 0V 7V/',# 0 TZTZZX S RTRRR UWTRYS S RTRRR 811-!' 2'-, <-$: *'"* 1#1 Y[ [T

~~185~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

~~! "~~

Figure 4.2 EGFR expression Immunostaining for EGFR was carried out using a mouse monoclonal antibody (Clone31G7, Zymed, CA, USA) which recognises the extracellular domain of the EGF receptor. The above shows a typical core from each of the TMA master blocks A-D. The top left hand core (a) is a leiomyosarcoma (LMS) from TMA master block A, showing membranous staining in >50% of the core. The top right hand core (b) is from master block B, depicting a high grade liposarcoma (LPS) with stronger membranous and cytoplasmic staining. The lower left hand core (c) is of a high grade spindle cell sarcoma, again with strong membranous and cytoplasmic staining of >50% of the core. The lower right hand specimen (d) is a synovial sarcoma with intense membranous staining throughout the core.

~~186~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

~~*-!)8 *-!)* ! "~~

~~*-!) *-!)~~

Figure 4.3 Activated EGFR expression The above is a selection of cores stained for the phosphorylated (activated) EGF receptor using a mouse monoclonal antibody (Chemicon International, Inc., Temecula, CA). This antibody recognises the intracellular domain of EGFR, resulting in cytoplasmic staining. The top left hand core (a) shows a low grade LMS, while the top right hand specimen (b) shows stronger staining in a high grade spindle cell sarcoma. A low grade LPS which did not express activated EGFR is shown in (c) but staining is present in the high grade LPS specimen (d).

~~187~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

~~/-5%0 "#/. *-!) *-!) %'%&%0 "#/.~~

~~! "~~

/-5%0 "#/ *-!)8 *-!)8 %'%&%0 "#/ Figure 4.4 Expression of p44/42MAPK (Erk1/Erk2) This rabbit polyclonal antibody (Cell Signaling Technology, Inc., Danvers, MA, USA) detects activated levels of Erk1 and Erk2 protein. The top left hand core (a) shows a low grade well differentiated LPS from the posterior thigh with lipoma like areas and few lipoblasts, while the top right hand core (b) shows a high grade de-differentiated retroperitoneal LPS, more cellular and with mitotic figures seen at higher magnification. Strong cytoplasmic staining is evident in the latter. A low grade LMS excised from the anterior chest wall of a patient is shown in (c), with elongated spindle cells, elongated nuclei but no prominent nucleoli. The cells form fascicles which infiltrate between collagen bundles. Strong cytoplasmic immunopositivity is seen. (d) depicts a high grade retroperitoneal LMS. This is a spindle cell sarcoma, fascicular in nature. Nuclear pleomorphism is present. There is strong nuclear and cytoplasmic staining shown in this core for p44/42MAPK.

~~188~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

~~! "~~

Figure 4.5 Phospho-Akt Expression This rabbit polyclonal antibody detects Akt 1 when phosphorylated at serine 473 and Akt2 and Akt3 only when phosphorylated at equivalent sites. Activation causes translocation from the cytoplasm to the nucleus. (a)High grade LMS, with plump spindle cells with large irregular pleomorphic nuclei, forming short fascicles. There is cytoplasmic and nuclear staining in this specimen. (b)Metastatic LMS of the lung with moderate pleomorphism, also showing nuclear and cytoplasmic staining. (c) Low grade myxoid LPS (d)Spindle Cell Sarcoma, with areas of fibro-, chondro- and osteosarcomatous differentiation. There is nuclear and cytoplasmic staining in this specimen. (e) Peripheral neuroextodermal tumour (PNET) in an adult. A highly cellular tumour with irregular nuclei and inconspicuous nucleoli, showing areas of cytoplasmic staining for pAkt.

~~189~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

~~4.3.6 pAkt~~

Akt or protein kinase B is an intermediate molecule in the PI3K-Akt signal transduction cascade (Figure 4.1 and Section 4.2.5.1.2). The rabbit polyclonal antibody detects Akt1 when phosphorylated at serine 473 and Akt2 and Akt3 only when phosphorylated at equivalent sites. Activation by receptor tyrosine kinases such as EGFR causes translocation from the cytoplasm to the nucleus. Strong immunopositivity for pAkt is manifest as both cytoplasmic and nuclear staining (Figures 4.5 and 4.7 to 4.9).

Table 4.14 Phosphorylated Akt in all STS samples across Histologic Grade .8)2 R S T 3-2 * %'12-*-%'! *-5%0 "# [ Y [ TW &0 "# ',2#0+#"' 2#%0 "# S T W Z &'%&%0 "# R U VT VW +#2 12 2'!%0 "# R S SR SS 3-2 * SR SU XX Z[

14.6% of all tumours demonstrated 1+ immunopositivity and 74.2% were scored as 2+. 88.9% of intermediate grade tumours expressed pAkt together with 100% of high grade and metastatic tumours. 36% of low grade tumours did not express pAkt. The correlation with tumour grade was statistically significant (Tables 4.20 and 4.21) The expression of phosphorylated Akt was compared to that of EGFR as well as activated EGFR, the raw data for which are given in Tables 4.8 and 4.10. The statistical correlations between pAkt and EGFR and pAkt and activated EGFR expression were significant on chi-square testing for all tumours regardless of neoadjuvant treatment (Tables 4.15 and 4.16).

~~190~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

Table 4.15 Correlation between EGFR and pAkt Expression (#- "(34 ,2 1#16!*3"#" "** 1#1 817+.T '%T 817+.T'%T ,".")2 : *3# "$ TV1'"#" : *3# "$ TV1'"#" .# 01-,*&'V/3 0# UWTYUU V RTRRR UYTVU[ V RTRRR /')#*'&--"( 2'- T[TRRR V RTRRR USTVRT V RTRRR /',# 0V 7V/',# 0 TRTWWT S RTRRR TSTTXU S RTRRR 811-!' 2'-, <-$: *'"* 1#1 YZ [S

Table 4.16 Correlation between Activated EGFR and pAkt Expression (#- "(34 ,2 1#16!*3"#" "** 1#1 "!2'4 2#" 817+.T '%T 817+.T'%T ,".")2 : *3# "$ TV1'"#" : *3# "$ TV1'"#" .# 01-,*&'V/3 0# VTTTVR V RTRRR VXTYWV V RTRRR /')#*'&--"( 2'- UWTRWU V RTRRR UYTVYR V RTRRR /',# 0V 7V/',# 0 TTTWX[ S RTRRR T[TUU[ S RTRRR 811-!' 2'-, <-$: *'"* 1#1 YZ [T

~~4.3.7 pSTAT3~~

The rabbit monoclonal antibody used in this study detects STAT3 which is phosphorylated at tyrosine 705. STAT3 is a key signal transduction molecule in the JAK-STAT pathway (Figure 4.1). This pathway may be triggered by growth factor receptor activation, such as EGFR. In addition, constitutive activation of STAT3 in certain tumours has also been noted (Section 4.2.5.1.3).

As for p44/42MAPK and pAkt, activation by phosphorylation results in cytoplasmic to nuclear translocation. Immunopositivity for the phosphorylated form of STAT3 is therefore manifest as nuclear staining (Figures 4.6 and 4.9). In general, far fewer tumours exhibited immunopositivity for this marker than for the others. Positive expression, when it was seen, tended to be focal and nuclear in nature. 73.6% of tumours did not express pSTAT3, while 21.8% demonstrated 1+ immunopositivity and only 4.6% were scored as 2+. 76% of low grade tumours did not express pSTAT3, while 87.5% of intermediate grade tumours did not express pSTAT3, together with

~~191~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

68.9% of high grade and 77.8% of metastatic tumours. The correlation with tumour grade was not statistically significant, as shown in Tables 4.20 and 4.21. This applied to the whole cohort, as well to the cohort who did not receive neoadjuvant treatment.

Table 4.17 pSTAT3 Immunoreactivity in all STS samples across Tumour Grade . "U R S T 3-2 * %'12-*-%'! *-5%0 "# S[ X R TW &0 "# ',2#0+#"' 2#%0 "# Y S R Z &'%&%0 "# UR SR V VV +#2 12 2'!%0 "#ZTRSR 3-2 * XV S[ V ZY

Phosphorylated STAT3 expression was examined in the context of EGFR and activated EGFR expression. The raw data is presented in Tables 4.8 and 4.10. Unlike the other activated signal transduction molecules investigated, the correlation between EGFR, activated EGFR and expression of pSTAT3 failed to reach statistical significance (Tables 4.18 and 4.19). Table 4.18 Correlation between EGFR and pSTAT3 Expression (#- "(34 ,2 1#16!*3"#" "** 1#1 817+.T '%T 817+.T '%T ,". "U : *3# "$ TV1'"#" : *3# "$ TV1'"#" .# 01-,*&'V/3 0# STVRV V RTZVU TTTUY V RTX[T /')#*'&--"( 2'- STY[S V RTYYV UTRZW V RTWVV /',# 0V 7V/',# 0 RTZZ[ S RTUVX STRXX S RTURT 811-!' 2'-, <-$: *'"* 1#1 YY [R

Table 4.19 Correlation between Activated EGFR and pSTAT3 Expression (#- "(34 ,2 1#16!*3"#" "** 1#1 "!2'4 2#" 817+.T '%T 817+.T '%T ,". "U : *3# "$ TV1'"#" : *3# "$ TV1'"#" .# 01-,*&'V/3 0# TTVXY V RTXWS ST[RS V RTYWV /')#*'&--"( 2'- UTSRS V RTWVS TTVYT V RTXWR /',# 0V 7V/',# 0 STRRT S RTUSY STYUY S RTSZZ 811-!' 2'-, <-$: *'"* 1#1 YY [R

~~192~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

~~! "~~

Figure 4.6 Phospho-Stat3 Expression Stat3 is activated by phosphorylation, which induces dimerisation, nuclear translocation and DNA- binding. pSTAT3(Tyr705) rabbit monoclonal antibody (Cell Signaling Technology, Inc., Danvers, MA, USA) detects endogenous levels of Stat3 only when phosphorylated at tyrosine 705. It does not cross-react with phospho-EGFR. (a)There is no immunopositivity in this specimen of a metastatic LMS. (b) Similarly, this myxoid LPS does not express activated Stat3. (c) There is focal nuclear staining in this myxofibrosarcoma (myxoid malignant fibrous histiocytoma). (d) Nuclear localisation of Stat3 is demonstrated in this core taken from a high grade epithelioid angiosarcoma.

~~193~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

~~-&"(~~

~~.8)2 ! " .VV VT8.7~~

~~.383U~~

Figure 4.7 EGFR and Signal Transducer Molecule Expression in Rhabdomyosarcoma The expression of the above proteins is examined in the same specimen. (a) There is strong membranous staining for EGFR (Clone31G7, Zymed, CA, USA). (b) Nuclear and cytoplasmic staining is noted with pAkt (Cell Signaling Technology, Inc., Danvers, MA, USA). (c) There is less intense cytoplasmic immunoreactivity for p44/42MAPK (Cell Signaling Technology, Inc., Danvers, MA, USA), with some focal nuclear localisation at the peripheral regions of the core (see insert). (d) This tumour did not express phosphorylated STAT3.

~~194~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

~~2 ~~

~~.VV VT#"' ! " .")2~~

Figure 4.8 H&E staining, EGFR and Signal Transducer Expression in Angiosarcoma The expression of the above proteins is examined in the same specimen. (a) The haematoxylin and eosin stained specimen is shown. (b) There is strong membranous staining for EGFR (Clone31G7, Zymed, CA, USA) at the vascular structures. (c) Focal cytoplasmic staining is noted with p44/42MAPK (Cell Signaling Technology, Inc., Danvers, MA, USA), while (d) more intense cytoplasmic and nuclear immunoreactivity for pAkt (Cell Signaling Technology, Inc., Danvers, MA, USA) is noted.

~~195~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

~~"!2'4 2#" .VV VT#"' ! "~~

~~.")2 . "U~~

Figure 4.9 Activated EGFR and Signal Transducer Expression in a Pleomorphic Sarcoma (PMS) The expression of the above proteins is examined in the same specimen. Pleomorphic tumour cells, frequent multinucleate forms, giant bizarre cells and scattered mitoses characterised this tumour, with a focal storiform pattern. There were extensive secondary changes such as necrosis, foamy macrophages, myxoid intimal vascular change, representing neoadjuvant treatment. However, abundant pleomorphic tumour was still present. (a) There is diffuse cytoplasmic staining for activated or phosphorylated EGFR (Chemicon International, Inc., Temecula, CA, USA). (b) There is cytoplasmic and nuclear immunoreactivity for p44/42MAPK (Cell Signaling Technology, Inc., Danvers, MA, USA. (c) Nuclear and cytoplasmic staining is noted with pAkt (Cell Signaling Technology, Inc., Danvers, MA, USA). (d) This tumour showed focal nuclear staining for phosphorylated STAT3.

~~196~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

~~4.3.8 Correlation between Biomarker Expression and Tumour Grade~~

Statistical correlation between the expression of EGFR, activated EGFR, the other activated downstream signal transduction molecules and histologic grade of the STS samples was evaluated in a number of ways.

~~4.3.8.1 Kruskal Wallis test~~

The Kruskal Wallis test, which can be applied to nonparametric data, is shown in Table 4.20. The biomarkers were tested in two ways, according to the immunostaining score assigned which ranged from 0-2, or more simply as positive or negative expression. The number 1 after each marker in the table refers to the 0-2 scoring system. Apart from pSTAT3, the expression of all the other biomarkers exhibited statistically significant correlation with tumour grade. This pattern of correlation applied to both the whole patient cohort, as well as to the restricted cohort comprising those who had not received neoadjuvant treatment.

~~4.3.8.2 Chi-Square tests~~

Pearson’s Chi-square tests were also applied to assess correlation between the expression of the proteins examined in this study and histologic grade. These results were then reconfirmed using Likelihood ratios and Linear-by-Linear associations. These are presented in Table 4.21. Again, apart from pSTAT3, the expression of all other biomarkers correlated with tumour grade. This was also true for the restricted cohort where patients who had received neoadjuvant treatment were excluded.

~~197~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

Table 4.20 Kruskal Wallis: Correlation between Grade and Biomarker Expression (#- "(34 ,2 1#16!*3"#" "** 1#1 0 "# ( ## , ,) . ( ## , ,) . *-5%0 "# TU URTTX RTRSYUZY TV UTTYW RTRSX ',2#0+#"' 2#%0 "# Z VTTS[ Z VXTYW &'%&%0 "# UY VTTZV VX VXTWT +#2 12 2'!%0 "# [ VTTYT X VW 3-2 * YY ZV S *-5%0 "# TU TWTU[ RTRRSW[V TV TYTUU RTRRS ',2#0+#"' 2#%0 "# Z UYTZZ Z VUTXU &'%&%0 "# UY VWTYV VX WRTUZ +#2 12 2'!%0 "# [ VYTRX X VSTTW 3-2 * YY ZV *-5%0 "# TU TXTW[ YT[W-VRX TV T[TTW R ',2#0+#"' 2#%0 "# Z VRTS[ Z VUTTW &'%&%0 "# UY VWTRR VX VZTW +#2 12 2'!%0 "# [ VWTRR X VZTW 3-2 * YY ZV S *-5%0 "# TU TXTZU RTRRVWS TV TYTZU RTRRS ',2#0+#"' 2#%0 "# Z UXTS[ Z UYTW &'%&%0 "# UY VWTZT VX V[T[Z +#2 12 2'!%0 "# [ VVTWX X WRTW 3-2 * YY ZV .")2 *-5%0 "# TU TZTXU RTRRRSW[ TV USTVV R ',2#0+#"' 2#%0 "# Z UZTYW Z VSTZS &'%&%0 "# UX VUTWR VW VY +#2 12 2'!%0 "# [ VUTWR X VY 3-2 * YX ZU .")2S *-5%0 "# TU TUTRY [TZT-VRY TV TVTZW R ',2#0+#"' 2#%0 "# Z UVTZZ Z UZTUZ &'%&%0 "# UX VYTWZ VW WSTRY +#2 12 2'!%0 "# [ VVTZU X VYTVT 3-2 * YX ZU .#"' *-5%0 "# TU URTSS RTRRRYYV TV UU R ',2#0+#"' 2#%0 "# Z VUTWR Z VY &'%&%0 "# UY VTTVX VX VXTR[ +#2 12 2'!%0 "# [ VUTWR X VY 3-2 * YY ZV .#"'S *-5%0 "# TU TTTWT STRT-VRX TV TUTTY R ',2#0+#"' 2#%0 "# Z UYTTW Z U[TVV &'%&%0 "# UY VYTZX VX WSTXU +#2 12 2'!%0 "# [ VXTTT X WUTW 3-2 * YY ZV . "U *-5%0 "# TU UYTSW RTVYYTRT TV VRTXU RTYWU ',2#0+#"' 2#%0 "# Z UUTX[ Z UWTWX &'%&%0 "# UW VRTY[ VV VTTYT +#2 12 2'!%0 "# [ UUTSY X VV 3-2 * YW ZT . "US *-5%0 "# TU UXT[U RTVUVYV TV VRTTW RTYSU ',2#0+#"' 2#%0 "# Z UUTWX Z UWTUZ &'%&%0 "# UW VRT[[ VV VUTRT +#2 12 2'!%0 "# [ UUTRX X VUTW 3-2 * YW ZT &#&'% *3#-$2� ,)$-0 %'4#,%0 "#Q2&#%0# 2#0'21 11-!' 2'-,5'2&2&##6.0#11'-,-$2&# '-+ 0)#0T&#bWc $2#0 2&# '-+ 0)#0 "#,-2#1 2&# '++3,-12 ',',%1!-0#-$ V 2- X 31#" 2- 120 2'$7 2&# '-+ 0)#0#6.0#11'-,T .�# 2�# '1 ,- bWc $-**-5',%2&# '-+ 0)#0Q2&##6.0#11'-,5 1120 2'$'#" 1.-1'2'4#-0,#% 2'4#-,*7T

~~198~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

~~Table 4.21 Chi-Square tests Correlating Biomarker Expression and Grade~~

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

~~199~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

~~4.3.9 Biomarker Expression and Histologic Subtype~~

The expression of each marker was further examined in the context of each histologic subtype. These were divided as described in Section 4.3.1. A Kruskal Wallis test was carried out, grouping the biomarker expression both according to positive/negative staining and the 3 point immunostaining score (0 to 2). The Kruskal Wallis test for nonparametric data assigns a rank for each of the histologic subgroups examined. The higher the rank for a given histologic group, the greater its association with the expression of the biomarker.

For EGFR expression, synovial sarcomas (SS) demonstrated the highest rank, followed by malignant fibrous histiocytomas/pleomorphic sarcomas (MFH/PMS). Leiomyosarcomas (LMS) and SS shared the highest rank for activated EGFR expression. The highest rank for pAkt expression was shown by the MFH/PMS group of tumours, followed by LMS. For p44/42MAPK, the highest rank was accorded to the LMS group of tumours, followed by SS. The greatest expression for pSTAT3 was seen in the last histologic grouping of “others”, which included rhabdomyosarcomas and angiosarcomas. These correlations were essentially the same for the two groups analysed, that is, the whole patient cohort and the group where those that received neoadjuvant treatment were excluded.

~~200~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

Table 4.22 Kruskal Wallis test for Biomarker Expression and Histologic Subtype (#- "(34 ,2 1#16!*3"#" "** 1#1 &'-+ 0)#0 2'12-*-%7 ( ,) . ( ,) . /. TT USTZV RTRWUWYV TT UYTSZ RTRYS[V[ / ST VVTYS SX WRTTW "% . TR VVTRW TV WRTTW W VZTRR X WXTRR -2 TR VRTSR TV VXTVT 3-2 * Y[ [T S /. TT TZTRY RTRTZVTX TT UUTUX RTRU[WX[ / ST VRTZZ SX VXTRR "% . TR VYTZZ TV WUTXR W VZTVR X W[TZU -2 TR VTTXU TV VZTVV 3-2 * Y[ [T /. TT USTXV RTRTY[VT TT UXTW[ RTRSSZRT / ST VTTYS SY WRTYX "% . TR VVTRU TV WSTWX W VXTRR X WUTWR -2 TR VTTRW TV VYTX[ 3-2 * Y[ [U S /. TT USTVS RTTRWYVX TT UWTZX RTRYVUWZ / ST VYTSY SY WXTVS "% . TR VTTTW TV WRTTU W VXTXR X WXTSY -2 TR VSTTW TV VWTRT 3-2 * Y[ [U .")2 /. TT TZTWW RTRRRTRY TT UTTXZ ST[-VRW / ST VSTTW SY VZTY[ "% . S[ VVTWR TU WSTWR W VVTWR X WSTWR -2 TR VVTWR TV WSTWR 3-2 * YZ [T .")2S /. TT TWTXS RTRRRUZS TT T[TXX RTRRRVUU / ST VUTVX SY WSTRU "% . S[ VXTVT TU WWTVU W UXTVR X VWTUU -2 TR VXTXR TV WRTVX 3-2 * YZ [T .#"' /. TT UUTYU RTRYYT[S TT UZTVW RTRTZZTS / ST VVTWR SX WSTRR "% . TR VTTWU TV V[TRZ W VVTWR X WSTRR -2 TR VRTWW TV VYTSY 3-2 * Y[ [T .#"'S /. TT T[TTY RTRTTRYS TT UUTUT RTRR[XXS / ST VZTWZ SX WXT[S "% . TR VTTSZ TV VZTWZ W VVTWR X WTTWZ -2 TR VUTUW TV VZTRV 3-2 * Y[ [T . "U /. TT VSTTW RTTRRZRW TT VYTUT RTTXUSRV / SS UXTRR SW VTTRR "% . TR UTTZW TV VRTWR W UXTYR X VRTWR -2 S[ VWTTS TU WTTWY 3-2 * YY [R . "US /. TT VRT[U RTSXVSRX TT VXTXZ RTTS[RRX / SS UWTZT SW VSTXR "% . TR UTTYW TV VRTX[ W UXTWR X VRTSY -2 S[ VWTZV TU WUTUU 3-2 * YY [R ) *V. **'1'1 ,-,. 0 +#20'!2#122& 2 1'%,1 0 ,)$-0# !&%0-3.T&#&'% *3#-$2&'10 ,) $-0 %'4#,&'12-*-%'!13 27.#Q2&#%0# 2#0'21 11-!' 2'-,5'2&2&##6.0#11'-,-$2&# '-+ 0)#0T

~~201~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

~~4.3.10 Disease Free Survival~~

Disease free survival was examined using chi square tests. Clinical variables as well as the expression of biomarkers were examined. Of the clinical factors, tumour site, grade, size and stage of disease were found to be significantly predictive (Table IV.7, Appendix IV). Other factors such as histologic subtype, resection margins and tumour depth were not significant. In considering the expression of the biomarkers examined, EGFR, its activated form and pAkt over-expression were found to significantly correlate with disease free survival for both groups of patients (Table 4.23 (1)). When the 3 point scoring system was used, p44/42MAPK expression was also found to be statistically significant. When patients who had received neoadjuvant treatment were excluded and the 3 point scoring system compared, pAkt and activated EGFR did not correlate with disease free survival. The possible reasons for this finding are considered in Section

~~202~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

~~Table 4.23 Correlation of Disease Free Survival and Biomarker Expression (1)~~

~~(#- "(34 ,26!*3"#" "** 2'#,21 + 0' *# + *3# . + *3# .~~

-&"( .# 01-,*&'V/3 0# VT[YS RTRTX VTR[S RTRVU *-,2',3'27*-00#!2'-, UTYVT RTRWU UTR[S RTRY[ /')#*'&--"( 2'- WTXVT RTRSZ VTUXV RTRUY /',# 0V 7V/',# 0811-!' 2'-, VT[RT RTRTY VTRVU RTRVV <-$: *'"* 1#1 YT ZW

-&"( .# 01-,*&'V/3 0# VTSR[ RTRVU XTTTU RTRSU *-,2',3'27*-00#!2'-, TTZVU RTR[T VTYVV RTRT[ /')#*'&--"( 2'- VT[ST RTRTY YTVUU RTRRX /',# 0V 7V/',# 0811-!' 2'-, VTRWT RTRVV XTSV[ RTRSU <-$: *'"* 1#1 YT ZW

.8)2 .# 01-,*&'V/3 0# UTWWV RTRW[ VTXZU RTRUR *-,2',3'27*-00#!2'-, TTUVV RTSTX UTUTS RTRXZ /')#*'&--"( 2'- VTTR[ RTRVR WTWRV RTRS[ /',# 0V 7V/',# 0811-!' 2'-, UTWRV RTRXS VTXTY RTRUS <-$: *'"* 1#1 YS ZV

.8.7 .# 01-,*&'V/3 0# TTSYW RTSVR TT[[Z RTRZU *-,2',3'27*-00#!2'-, STSYX RTTYZ STZUV RTSYX /')#*'&--"( 2'- TTWTR RTSST UTVWW RTRXU /',# 0V 7V/',# 0811-!' 2'-, TTSVW RTSVU TT[XU RTRZW <-$: *'"* 1#1 YT ZW

.383U .# 01-,*&'V/3 0# TTTYU RTSUT UTWVY RTRXR *-,2',3'27*-00#!2'-, STVW[ RTTTY TTXVT RTSRV /')#*'&--"( 2'- TTS[X RTSUZ UTV[T RTRXT /',# 0V 7V/',# 0811-!' 2'-, TTTVS RTSUV UTWRV RTRXS <-$: *'"* 1#1 YR ZU 3&#,3+ #0S $2#02&# '-+ 0)#00#$#012-2&#U.-',21!-0',%1712#+RVT$-0 '++3,-12 ',',%1##,#623 *# <-,3+ #0 $2#02&# '-+ 0)#00#$#012-2&#!-&-02"'4'"#"', ', 07$ 1&'-,Q 1.-1'2'4# -0,#% 2'4#12 ',',%

~~203~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

Disease Free Survival and Biomarker Expression (2) (#- "(34 ,26!*3"#" "** 2'#,21 + 0' *# + *3# . + *3# . S .# 01-,*&'V/3 0# ZTUZU RTRSW XTWRT RTRU[ /')#*'&--"( 2'- ZT[SX RTRST XTYTY RTRUW /',# 0V 7V/',# 0 ZTSY[ RTRRV XTVRW RTRSS 811-!' 2'-, <-$: *'"* 1#1 YT ZW

~~S .# 01-,*&'V/3 0# VTSTX RTSTY XTYW[ RTRUV /')#*'&--"( 2'- VT[T[ RTRZW YT[WY RTRS[ /',# 0V 7V/',# 0 TTYWZ RTR[Y WTXZT RTRSY 811-!' 2'-, <-$: *'"* 1#1 YT ZW~~

~~.")2S .# 01-,*&'V/3 0# UTXRZ RTSXW VTXZV RTR[X /')#*'&--"( 2'- VTTXS RTSS[ WTWRV RTRXV /',# 0V 7V/',# 0 TTVVR RTSSZ UTUWY RTRXY 811-!' 2'-, <-$: *'"* 1#1 YS ZV~~

.VV VT#"'S .# 01-,*&'V/3 0# SRTRZ[ RTRRX SXTS[U RTRRR /')#*'&--"( 2'- SSTYSX RTRRU SZT[XU RTRRR /',# 0V 7V/',# 0 YTZT[ RTRRW STTXVR RTRRR 811-!' 2'-, <-$: *'"* 1#1 YT ZW . "US .# 01-,*&'V/3 0# TTTYU RTUTS VTTRW RTSTT /')#*'&--"( 2'- TTS[X RTUUV VTSXY RTSTV /',# 0V 7V/',# 0 TTRSW RTSWX VTSVY RTRVT 811-!' 2'-, <-$: *'"* 1#1 YR ZU 3&#,3+ #0S $2#02&# '-+ 0)#00#$#012-2&#U.-',21!-0',%1712#+RVT $-0'++3,-12 ',',%T -&"(S8!2'4 2#"-&"(

~~204~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

~~4.3.11 Overall Survival~~

Univariate and multivariate Cox regression analyses were carried out, the latter using the forward likelihood ratio method. Variables included histologic subtype, tumour size, site, depth, stage, resection margins and grade, as well as expression of all biomarkers investigated. Of the biomarkers, only activated EGFR and phosphorylated Akt were found to be independent predictors of overall survival, with p < 0.05 on multivariate analysis (Table 4.24). Tumour grade and stage also reached statistical significance, when the whole patient cohort was analysed. In the restricted group where patients who had received neoadjuvant treatment were excluded, tumour grade failed to reach statistical significance, but activated EGFR, pAkt and stage remained independent predictors of outcome (Table 4.25). As noted in the previous section however, tumour grade was statistically significant for this group when disease free survival was considered (Table IV.7, Appendix IV). The variables that failed to reach significance on multivariate analysis are shown in Table 4.26 as part of the univariate analysis. Cancer specific survival was also recorded. However, as there was only one patient in this study for whom the cause of death was unknown, the remainder having died as a result of their STS, overall survival was used as the main outcome measure for the purpose of this analysis.

4.24 Multivariate Cox Regression Analysis for Independent Predictors of Overall Survival #3*2'4 0' 2#", *71'1$-0"** 2'#,21 [WTR$*$-0%( : 0' *# . %( /-5#0 >..#0 -&"( RTRRR TTTTZ STVYS UTUYX .8)2 RTRRU TTRUT STTYV UTTVY &0 "# RTRT[ STTYW STRTW STWZY 2 %# RTRU[ STUTW STRSW STYUS -&"(S 8!2'4 2#" -&"( U .-',2 1!-0',% 1712#+R .8)2S .&-1.&-07* 2#" 8)2 U .-',2 1!-0',% 1712#+R%(S% 8 0"( 2'-R*S*-,$'"#,!#,2#04 *R&0 "#SV&0 "#1/-52-#2 12 2'!R2 %#SV :

~~205~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

~~Table 4.25 Multivariate Analysis for Overall Survival Excluding Neoadjuvant Cases~~

#3*2'4 0' 2#", *71'16!*3"',%(#- "(34 ,2 1#1 [WTR$*$-0%( : 0' *# . %( /-5#0 >..#0 -&"(S RTRRR TT[ZZ STYSZ WTS[W .8)2S RTRRU TTUYV STUVY VTSZV 2 %# RTRS[ STWXU STRYX TTTX[ /#S !2'4 2#"/#Y.-',21!-0',%1712#+R. )2S &-1.&-07* 2#" )2Y.-',21!-0',%1712#+R #S 8 0"# 2'-R$S$-,$'"#,!#,2#04 *R2 %#SV0

Table 4.26 Univariate Analysis for Overall Survival !,'4 0' 2#", *71'1$-0"** 2'#,21 [WTR$*$-0%( : 0' *# . %( /-5#0 >..#0 %'12-*-%7 RTZSX STRSY RTZZT STSYU '8# RTR[S STTVU RT[XX STW[[ #.2& RTZ[X STRUW RTXSX STYU[ '2# RT[SU RT[[T RTZWT STSWV 0%', RTUUV STSRS RT[RX STUU[ <#- "(34 ,2(6 RTTSY RTXYX RTUXU STTWZ -&"( RTUTV STSVS RTZYY STVZW .VV VT8.7 RTRUT STVWS STRUT TTRVR .383U RTUTW RTZS[ RTWWS STTS[

On univariate analysis, p44/42MAPK over-expression was associated with poor clinical outcome (HR = 1.451, p = 0.032), however it was not shown to be an independent predictor of survival on multivariate analysis. p44/42MAPK over-expression only showed a trend towards statistical significance on univariate analysis of the cohort where those who received neoadjuvant treatment were excluded.

The survival curves for activated EGFR, phosphorylated Akt as well as tumour grade and size were plotted (Figures 4.10 to 4.14) and log rank testing (Mantel-Cox) computed, as a further means of assessing statistical significance.

~~206~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

~~4.3.11.1 Tumour Grade and Survival~~

Cox regression analysis revealed tumour grade to be a significant predictor of survival. Clinical outcome was better in those with low grade tumours up to a follow up period of 65 months (Figure 4.10). However, the log rank test did not reach statistical significance. This may be a consequence of the length of follow up for the patients in this study or a result of the sample size. The STS samples used to construct the TMA included specimens which were resected as late as mid 2005. Thus the median follow up was 32 months and 70% of patients were followed up for only 50 months.

~~4.3.11.2 Disease Stage and Survival~~

The stage of the disease process was also an independent predictor of overall survival on multivariate analysis of the whole patient cohort (HR = 1.325, p = 0.039). These results remained statistically significant when patients who had undergone neoadjuvant treatment were excluded. As would be expected, survival was poor for those with advanced stages of malignancy. The Kaplan Meier curves are shown in Figure 4.11. Log rank testing was significant in this instance.

~~4.3.11.3 Tumour Size and Survival~~

The greatest difference in overall survival was seen between those with tumours ≤ 5cm and those with tumours > 10cm (Figure 4.12). However, Cox regression analysis had shown that tumour size was not an independent predictor of survival. The log rank test similarly failed to reach significance.

~~4.3.11.4 Activated EGFR and Survival~~

There were significant differences in survival according to degree of immunopositivity for activated EGFR (Figure 4.13). The best survival was seen in those who did not express activated EGFR (immunostaining score = 0). Conversely, the worst survival was seen in the group over-expressing activated EGFR (immunostaining score = 2+). The Log Rank test confirmed statistical significance with a p value of 0.05. The expression of activated EGFR therefore correlated with tumour grade and was found to be an independent predictor of survival (HR = 2.228, p <0.001). In effect, activated EGFR was shown to be a better prognostic marker than conventional prognostic indicators such as resection margins, tumour size or grade.

~~207~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

~~4.3.11.5 Phosphorylated Akt and Survival~~

Phosphorylated Akt was also determined by Cox regression analysis to be a statistically significant independent predictor of overall survival (HR = 2.032, p = 0.003) when the whole cohort was analysed. When patients who received neoadjuvant therapy were excluded, the results remained significant. The Kaplan Meier survival curve was plotted for pAkt expression, stratified according to positive and negative expression (Figure 4.14). Better initial survival is seen in the group not expressing pAkt. The curves then intersect and the log rank test failed to reach significance. The small sample size in this study may have contributed to this discrepancy. It is also possible that pAkt exerts a transient effect during the early proliferative phase of tumour growth. Hence the influence on overall survival may not be sustained.

~~208~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

~~/-%0 ,)Q.1RTRW~~

Figure 4.10 Overall Survival for Patients Stratified by Histologic Grade (2 groups) Tumour grade was found to be significant on multivariate analysis (p = 0.029). When a survival analysis for tumour grade was carried out grouping the high grade and metastatic tumours together, the Kaplan Meier curves showed a clear separation, with better survival for those with low grade tumours. However, the difference was not significant on log rank testing, probably due to the short period of follow up for many of these patients.

~~209~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

Figure 4.11 Overall Survival for Cohort stratified by Stage of Disease The curves were plotted stratifying the patient cohort into two groups: those with Stage 1 and 2 disease in one group and those with Stage 3 and 4 disease in the other. The group with more advanced disease clearly had a poorer outcome. The log rank test for this stratification was significant, with a p value of 0.022.

~~210~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

~~/-%0 ,)Q.-RTSVX~~

Figure 4.12 Overall Survival for Patients Stratified by Tumour Size (2 groups) When a survival analysis for tumour size was carried out for tumours ≤≤≤ 5cm and those > 10cm, the Kaplan Meier curves showed a clear separation, with better survival for those with smaller tumours. However, the difference was not significant on either the cox regression analysis or log rank testing.

~~211~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

~~8**. 2'#,21S/-%0 ,)Q.-RTRW <#- "(34 ,2. 2'#,21#6!*3"#"S/-%0 ,)Q.-RTRU[~~

Figure 4.13 Overall Survival of Patients Stratified for Activated EGFR Expression The degree of immunopositivity for activated EGFR was scored from 0 to 2+ for the tissue microarray cores. On multivariate cox regression analysis, activated EGFR expression was found to be an independent predictor of overall survival (p = 0.002) in these patients. The Kaplan-Meier survival curves were plotted, as shown above. The best survival was seen for those patients who did not express activated EGFR. The worst survival was seen in those overexpressing activated EGFR (immunostaining score 2+). Importantly, the Log rank test also yielded a statistically significant p value of 0.05. The log rank test yielded a p value of 0.039 for the smaller cohort, where patients who had undergone neoadjuvant treatment were excluded.

~~212~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

~~/-%0 ,)Q.1RTRW~~

Figure 4.14 Overall Survival of Patients Stratified for Phosphorylated Akt (pAkt) Expression The survival curves were plotted with the immunopositivity for pAkt divided more simply into those with positive versus negative expression. On multivariate cox regression analysis, pAkt expression was found to be an independent predictor of overall survival (p = 0.003) in these patients. Those that did not express pAkt, appeared initially to have better survival rates. However, the Log rank test did not reach statistical significance. This may be due to the small sample size in the study. Alternatively, it is possible that Akt exerts only a transcient effect in these tumours, such as in the early proliferative phase. Hence at later stages, there appears to be no effect on overall survival.

~~213~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

~~4.4 DISCUSSION~~

Based on earlier findings on gene expression arrays and qRT-PCR, the aim of this chapter was to examine the protein expression of EGFR in clinical samples of STS using the technique of tissue microarray. In addition, whether EGFR played a functional role in tumour progression was assessed in part by evaluating the expression of activated EGFR and in part by correlation to clinical outcome. The downstream signal transduction pathway(s) that may be utilised by EGFR in STS were also investigated.

~~A number of limitations peculiar to the technique of tissue microarray and immunohistochemistry in general, were identified in relation to this project and these are considered in Appendix IV.~~

~~4.4.1 Correlation with Survival~~

The major findings were that activated EGFR and pAkt correlated with tumour grade and were independent predictors of survival. This clearly has implications not only for further functional studies, but also for potential use of EGFR- and Akt-directed targeted therapies in the treatment of the major types of STS and this is elaborated on further in Section 4.4.2 to 4.4.3.

~~4.4.1.1 Tumour Grade, Stage and Other Clinical Parameters~~

Tumour grade was found to predict overall survival on multivariate analysis for the whole patient cohort and disease free survival in the cohort where those who had undergone neoadjuvant treatment were excluded. However, this was not borne out on log rank testing, contrary to other published data(Markhede, Angervall et al. 1982; Trojani, Contesso et al. 1984; Coindre, Terrier et al. 2001; Kaytan, Yaman et al. 2003; Zagars, Ballo et al. 2003b). This unexpected finding can only be explained by the short duration of follow up in this patient cohort. Only 8 of the patients in this group underwent surgical resection of their tumours prior to 2000. Thus for a large number of the patients, the outcome was “censored” at the end of the study period, rather than continuing follow up until time of death.

~~The criteria employed for the grading of STS include factors such as nuclear pleomorphism, degree of necrosis and presence of mitotic figures. In other words, it~~

~~214~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

serves as a measure of the aggressiveness of the tumour and thus should intuitively be associated with clinical outcome. The patients who had undergone neoadjuvant treatment were initially analysed together with the rest of the patients. This was because their resected tumours were deemed by our independent pathologist to contain significant amounts of viable aggressive high grade tumour. The tumours were sampled from these regions for the tissue microarray. Thus on excluding these patients from the second analysis, it could be argued that 13 patients with high grade tumours were excluded, potentially skewing the results and further reducing the sample size.

~~The stage of the disease, which is a composite of the usual TNM criteria and tumour grade, was also an independent predictor of overall survival in both analyses, in keeping with published data.~~

While tumour size was found to correlate with disease free survival, it was not found to be an independent predictor of overall survival on multivariate cox regression analysis. Similarly, tumour depth failed to reach statistical significance. Studies which have proven these to be independent predictors of survival have generally had large sample sizes of > 500 patients, with longer follow up periods than was possible in our study. Moreover, our study population was skewed in that 73 of the 92 tumours were deep in location.

Our study did not find surgical resection margins to be predictive of survival. Again the study population was somewhat skewed in that 59 of the 92 resected tumours had margins that were closer than 1 mm, and in a further 7 cases, the margin was not clearly recorded. There have however been conflicting reports in the literature regarding the relationship of involved surgical margins with overall survival, although it is accepted that involved margins are a risk factor for the development of local recurrence. Only certain large studies have been able to show a statistically significant association with the development of metastases or survival (Pisters, Leung et al. 1996; Trovik, Bauer et al. 2000; Stojadinovic, Leung et al. 2002b).

~~215~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

~~4.4.1.2 Exclusion of Neoadjuvant Therapy Cases~~

All specimens selected for inclusion on the tissue microarray were examined by a trained pathologist at Prince of Wales Hospital. The thirteen specimens obtained from patients who had undergone neoadjuvant treatment were included as they were determined to have significant viable tumour present, typically still judged to be aggressive or high grade. The cores were sampled from these designated areas of high grade tumour. Thus it was not felt that these patients should be excluded from the analysis performed.

Time constraints limited the number of specimens that could be examined and selected by our pathologist and hence a decision was made to restrict our sample to patients operated on from 1999 to 2005. This unfortunately limited our sample size and may have contributed to the lack of statistically significant correlation between variables such as size and depth of tumour and survival, contrary to other much larger published studies examining clinical prognostic indicators. Similarly, the patient cohort included only 9 patients with local recurrences and separate analysis of this small subgroup was not carried out. These same issues of reduced sample size, short duration of follow up and affective exclusion of 13 patients judged by our pathologist to have significant viable high grade tumour could also explain the lack of correlation of pAkt (when using the 3 point scoring system) with disease free and overall survival.

~~4.4.1.3 p44/42 MAPK and pSTAT3~~

Phosphorylated MAPK expression was found to correlate with tumour grade, EGFR as well as activated EGFR. It was a predictor of survival on univariate analysis but was not significant on multivariate analysis. Phosphorylated STAT3, on the other hand, had very low overall expression in STS and did not correlate with the expression of EGFR or activated EGFR. A few possible reasons for the non correlation can be postulated: p44/42MAPK and pSTAT3 may be modulated by other growth factor receptors, cytokines, cell cycle regulators or mutations thereof. The relationship between EGFR activation and MAPK activation may be a direct one but other factors could then modulate its activity. An example in lung adenocarcinoma is the finding that Kras and EGFR mutations are mutually exclusive (Eberhard, Johnson et al. 2005).

~~216~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

There may be cross-talk between pathways. A conserved serine in the transcriptional activation domain (TAD) of STATs can be phosphorylated by other kinases such as Erks and p38 in the MAPK cascade, resulting in decreased transcriptional activity of the STAT (Calo, Migliavacca et al. 2003) and it is possible that this phenomenon is at play in STS. This may explain increased p44/42 MAPK expression but reduced pSTAT3 expression. The cross talk may involve cell cycle regulators. Cyclin D1’s (CCND1) CDK-independent activities include modifying gene transcription of signal transducers and hormones such as STAT3, C/EBPβ, BETA2 and oestrogen receptor (ER) (Lamb and Ewen 2003; Ewen and Lamb 2004; Fu, Wang et al. 2004; Knudsen, Diehl et al. 2006). CCND1 is postulated to exert an inhibitory effect on STAT3. CCND1 was shown in the gene expression arrays to be upregulated in the high grade tumour cell line. It has been shown in other studies to be overexpressed in a variable proportion of STS (Creager, Cohen et al. 2001; Kim, Cho et al. 2001; Yoo, Park et al. 2002). It may therefore be modulating STAT3 expression in these tumours. pMAPK in some studies have only been seen at the advancing margins or leading edge of the tumour. In other words, there may be heterogeneous distribution of the biomarker within the tumour. However, the immunostaining pattern for p44/42MAPK noted in this project was of a widespread, diffuse nature with a high prevalence among the higher tumour grades. Thus it seems unlikely that expression of p44/42MAPK in this group of tumours is restricted to the advancing edge. In the case of the focal nature of pSTAT3 staining, however, as discussed below, this may have resulted in under-reporting of its immunopositivity as only small cores were analysed.

~~4.4.2 Significance of Findings in this Study~~

~~The most significant findings of this study were that activated EGFR and phorphorylated Akt were shown to be independent predictors of survival.~~

We demonstrated a high overall prevalence of EGFR expression. These findings concur with a Japanese study which found an overall prevalence of 60% in their STS specimens (Sato, Wada et al. 2005). However, as with the Japanese study, our findings showed wild-type EGFR not to be an independent predictor of survival once multivariate cox

~~217~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

regression analysis was carried out. Another gene expression study of STS also identified EGFR as being overexpressed, mainly in synovial sarcomas (Baird, Davis et al. 2005). However, protein expression was only evaluated in a few samples, activated EGFR expression and signal transduction was not examined and no clinical correlation was carried out. Other earlier studies similarly had found EGFR overexpression mainly in synovial sarcomas (Nielsen, Hsu et al. 2003; Thomas, Giordano et al. 2005) rather than across all the major histologic subtypes.

The major novel finding of this project was a high rate of activated EGFR expression across the various histologic subtypes of STS examined. There was a strong positive correlation with histologic grade of malignancy. As tumour grade is a measure of the aggressiveness of malignant behaviour, it can be postulated that EGFR is functional and has a role in tumour progression in sarcoma. Of note, activated EGFR expression was found to be a prognostic marker, with overexpression correlating with poor survival on multivariate analysis. Thus the expression of this biomarker holds prognostic significance.

To our knowledge, EGFR signal transduction has not been investigated in soft tissue sarcomas. Two of the three activated signal transducers (p44/42MAPK and pAkt) examined, were also expressed in almost 90% of tumours. Both showed a strong positive correlation with histologic grade. In addition, pAkt was found to be a predictor of survival, with over-expression associated with a worse outcome. Its effect may, however, be transient, as suggested by the Kaplan Meier survival curves. These findings indicate that activated EGFR may exert its function through either or both the Ras- MAPK and PI3K-Akt pathway(s) in sarcoma.

Functional studies are a logical progression from the above and are, in fact underway in our laboratory. Animal knockout models, siRNA and preclinical drug trials on sarcoma cell lines are options to be considered and all of these should evaluate the effect on the signal transduction pathways noted above.

~~Evaluating the mechanism of activation of EGFR itself is another logical area of future study. Certain known ligands of EGFR such as epiregulin were shown in the gene~~

~~218~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

expression arrays to be overexpressed in the high grade sarcoma cell lines and would warrant further validation. EGFR is also known to be activated indirectly by stimuli such as oxidative and mechanical stress, hyperosmolarity and γ radiation.

EGFR transactivation, has recently been the subject of research in other tumour cell lines, wound healing and other physiologic processes also pose an area of interest. To my knowledge, this phenomenon is yet to be investigated in the context of sarcomas. Potential transactivators of EGFR include • G protein coupled receptors (GPCRs)27 via intermediaries such as MMPs28 and kinases (protein kinase C, src and JAK2) (Fischer, Hart et al. 2003; Barnes and Kumar 2004; Drube, Stirnweiss et al. 2006). GPCRs can in turn be activated by factors such as thrombin and endothelin-1, which are involved in cell migration in smooth muscle cells (Kalmes, Vesti et al. 2000);

• β2 adrenergic receptors, independent of MMPs, via extracellular Erk/MAPK (Drube, Stirnweiss et al. 2006); • Prostaglandin E2 (PGE2), via src kinase and MMPs (ADAM: a disintegrin and metalloprotease), which releases TGFα 29 from the cell membrane (Pai, Soreghan et al. 2002)30;

27 GPCRs are involved in functions as diverse as neurologic transmission, metabolism, growth, differentiation and cell migration (Fukuhara, Chikumi et al. 2001). In an early study, treatment of rat fibroblasts with thrombin and endothelin-1 resulted in EGFR transactivation by GPCRs (Daub, Weiss et al. 1996) 28 Metalloproteases involved in this signaling process are termed members of the ADAM (a disintegrin and metalloprotease) family of zinc-dependent proteases (Fischer, Hart et al. 2003) 29 TGFα is a known ligand for EGFR activation 30 This paper discussed the role of PGE2 and EGFR transactivation in the development of colon cancer and gastrointestinal hyperplasia. Extracellular Erk2 was thought to be involved in signal transduction. 219

~~Chapter 4: EGFR and Signal Transduction in STS~~

~~• Urokinase plasminogen activator and its receptor (uPA and uPAR), via the~~

activation of the cell adhesion molecule Integrin α5β1 and fibronectin, as demonstrated in Hep3 human carcinoma cells (Barnes and Kumar 2003); • TNFα coverting enzyme (TACE or ADAM17), which is involved in the processing of amphiregulin, an EGFR ligand, as shown in head and neck squamous cell carcinomas (Fischer, Hart et al. 2003) and • TGFβ, which has been shown in hepatoma cell lines to induce EGFR ligands TGFα and heparin-binding EGF (HB-EGF) (Caja, Ortiz et al. 2007). It is of note that TGFβ was overexpressed in the high grade sarcoma cell line on the gene expression arrays (Chapter 2), although this could not be validated on real time RT-PCR. Unpublished data from our laboratory, however, supports the findings of the expression arrays, as TGFβ protein over-expression was found to correlate with tumour grade in the 46 samples of STS studies. The dual role of TGFβ in suppressing and promoting carcinogenesis in epithelial tumours has been discussed (Chapter 2, Section 2.4.4.3.1 and Figure 2.15). The present study shows a potential tumour promoting effect in sarcoma. It is also possible, that TGFβ transactivation of EGFR is occurring in these tumours.

~~4.4.3 Targeted Therapy~~

The ultimate goals of translational research in the field of oncology are diagnostics and therapeutics. In the management of sarcoma, in particular, where 50% of patients still die from their disease, despite advances in surgery and radiotherapy, there is an acute need for identifying prognostic markers of tumour progression and metastasis and using these, where possible, as therapeutic targets.

The success of drugs such as imatinib (STI 571/Gleevec®/Glivec®, Novartis, Basel, Switzerland) in the treatment of gastrointestinal stromal tumours (GISTs) and Trastuzumab (Herceptin, Genentech Inc., South San Francisco, CA, USA) for breast cancers has spurred the search for other such treatments.

Drugs targeting EGFR have been in use for some years. Strategies include the use of monoclonal antibodies against EGFR’s extracellular domain, small molecule inhibitors of the tyrosine kinase domain that bind reversibly or irreversibly, natural inhibitors, as

~~220~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

~~well as synergistic combinations of drugs (Barnes and Kumar 2004; Baselga and Arteaga 2005; Bianco, Melisi et al. 2006).~~

By far the most studied of these drugs are the quinazolines Gefitinib (ZD1839/Iressa®) and Erlotinib (OSI-774/Tarceva®) which are tyrosine kinase inhibitors. Gefitinib and Erlotinib have been trialled in the treatment of lung, breast, renal, ovarian, prostate, head and neck and colorectal malignancies (Daneshmand, Parolin et al. 2003; Dawson, Guo et al. 2004; Perez-Soler, Chachoua et al. 2004; Baselga, Albanell et al. 2005; Siegel-Lakhai, Beijnen et al. 2005; Tsao, Sakurada et al. 2005b). They have been approved for use in non small cell lung cancer (NSCLC) and pancreatic cancer. Lessons have been learnt from these trials with regard to responsiveness of these tumours to EGFR inhibition.

In the case of NSCLC, EGFR overexpression alone has not been sufficient to predict sensitivity to the drugs. Activating mutations in the tyrosine kinase domain have been found in those that responded (Lynch, Bell et al. 2004; Paez, Janne et al. 2004). In gliomas, an amplified variant of the EGFR receptor conferred resistance to treatment in a proportion of patients (Haas-Kogan, Prados et al. 2005). In the same Phase 1 trial of 41 patients, 8 patients with and without EGFR amplification responded to Erlotinib. Interestingly, there was no response to treatment in any patient whose tumour exhibited positive expression of the downstream signal transducer pAkt.

It is not known what factors may predict responsiveness to EGFR inhibition in sarcoma. A recent study was able to demonstrate a lack of mutations in the EGFR gene (Baird, Davis et al. 2005) in the sarcomas studied. However, it cannot be assumed that mutations are required for responsiveness simply because this is the case in NSCLC. Indeed, a successful preclinical trial utilizing a combination of Erlotinib and Interferon- α on bladder cancer cell lines with no demonstrated EGFR mutations lends credence to this argument (Yang, Qu et al. 2007). This thesis has shown not only the expression of EGFR in a high proportion of sarcomas but also an over-expression of the activated (phosphorylated) form of EGFR and this may confer sensitivity. Based on these findings, the preclinical trials on sarcoma cell lines are currently being undertaken in our laboratory.

~~221~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

This thesis also confirmed a high rate of over-expression of phosphorylated Akt in sarcomas. It is possible that for these cases, a combined approach may be required, with inhibitors directed at both EGFR and Akt. Small molecule inhibitors of Akt are in the preclinical phase of development (Bianco, Melisi et al. 2006). Approximately 75% of STS in this study also demonstrated p44/42MAPK overexpression and small molecule inhibitors of MAPK kinase are being developed (Solit, Garraway et al. 2006).

To date, combination therapy in STS has included broader spectrum agents targeting, for instance, more than one tyrosine kinase. The anti-angiogenic agent Bevacizumab (Avastin™) was trialled in combination with Doxorubicin in 17 patients with metastatic STS, albeit with a poor response rate of 12%, similar to that achieved with doxorubicin alone (D'Adamo, Anderson et al. 2005). Sunitinib malate (SU11248), which targets the tyrosine kinases KIT, PDGFR, VEGFR and FLT3, has been trialled successfully in patients with GISTs refractory to Imatinib treatment (Demetri, van Oosterom et al. 2006). The raf kinase inhibitor Sorafenib (BAY 43-9006/Nexavar™), which targets raf as well as VEGFR is being used in a trials involving patients with advanced sarcoma (Hartmann 2007; Kasper, Gil et al. 2007).

It is likely, given the complexity of oncogenesis and tumour progression, that several factors are likely to be driving these processes, rather than a single one. It stands to reason, therefore, that an approach combining these small molecule inhibitors with each other or with conventional chemotherapeutic agents would have a better chance of success.

~~222~~

~~Chapter 4: EGFR and Signal Transduction in STS~~

~~4.5 Summary~~

• A tissue microarray of STS specimens was constructed in order to determine the protein expression of EGFR, which was found on gene expression arrays and real time RT-PCR to be over-expressed at the transcript level in the high grade and metastatic sarcoma cell line. • Over-expression of EGFR and activated EGFR was demonstrated in a large proportion of the tumours and there was a positive correlation with tumour grade, thus confirming one of the central hypotheses of this thesis. • Activated EGFR and pAkt expression were found to be independent predictors of survival on multivariate Cox regression analysis. In addition, pMAPK expression was significant on univariate analysis. Crucially, this confirms the EGFR pathway to be functionally active in sarcoma tumour progression and prognostically significant. • The above findings have significant implications for further functional studies of EGFR and the related pathways, as well as for the therapeutic targeting of EGFR for the treatment of soft tissue sarcomas.

~~223~~

~~Chapter 5: Characterisation of Cell Lines~~

~~2" +~~

~~5. ESTABLISHMENT and METHODS of CHARACTERISATION of NEW SARCOMA CELL LINES~~

~~224~~

~~Chapter 5: Characterisation of Cell Lines~~

~~5.1 INTRODUCTION~~

One of the specific aims of this project was to establish and characterise new sarcoma cell lines, due to the perceived need for more in vitro models for functional and preclinical studies. This was done concurrently with the remainder of the project, which dealt with gene and pathway discovery in the context of a tumour progression model in sarcoma. Numerous primary cultures were established from STS resected at the Prince of Wales Hospital between November 2003 and March 2005. These were subcultured and their morphology and immunocytochemistry profile examined in order to compare the cultures to the original source tumours. Further characterisation was carried out on two of the cell lines. The establishment of the short term cell cultures is described in this chapter, together with the methods used for further characterisation of the new cell lines, GIST-M and LMS-LFS. The karyotyping was kindly performed by Dr Robyn Lukeis at SydPath, St Vincent’s Hospital. The TP53 and KIT mutational analyses were carried out in collaboration with Professor Rodney Scott at Hunter Area Pathology Service and Drs Maurice Loughrey and Victoria Beshay at Peter MacCallum Cancer Centre respectively. The telomerase repeat amplification protocol (TRAP) assay and ALT-associated promyelocytic leukaemia bodies (APB) staining was kindly performed by Dr Jeremy Henson at the Children’s Medical Research Institute, Westmead.

~~5.1.1 Cell lines as resources~~

One of the major problems in the management of STS, with the exception of a few STS, such as the paediatric sarcomas, is the resistance to traditional chemotherapeutic agents. Given that mortality occurs largely as a result of metastatic disease, new systemic therapies must be developed to address this unmet medical need.

Important resources in developing our understanding of the molecular biology of STS are the in vitro and xenograft models we establish in the laboratory. Cell lines, that can be grown indefinitely, provide large amounts of homogeneous material for experimentation including the assessment of potential therapeutic agents. Mechanisms of drug action can be assessed on in vitro models, as can the effects of drug treatment on specific tumour cell characteristics such as cell proliferation, apoptosis and invasion.

~~225~~

~~Chapter 5: Characterisation of Cell Lines~~

Cell lines have also been used in gene expression profiling studies to identify the gene signatures associated with STS subtypes. As discussed in Chapter 1 (Section 1.10.3.2.1), these have typically involved sarcomas that are more common in the paediatric population, such as Ewing’s sarcoma (ES), alveolar rhabdomyosarcoma (ARM) and the histologically similar tumours grouped together as the small round blue cell tumours (Khan, Simon et al. 1998; Schaefer, Wai et al. 2002; Wai, Schaefer et al. 2002; Schaefer, Brachwitz et al. 2004; Schaefer, Brachwitz et al. 2006). This is in contrast to the common adult STS such as leiomyosarcomas (LMS), where relatively few cell lines have been established and characterised.

In addition to the relative paucity of adult non gynaecologic STS cell lines, for the cell lines that do exist, associated relevant data such as the histopathology of the tumour of origin and the extent to which the cell line retains the characteristics of the tumour, is not always available. This is partly because many of these cell lines were derived 20 to 30 years ago. The majority of cell lines from adult STS were derived from highly aggressive cancers; the low grade cancers are under-represented. In a review of sarcoma cell lines, Beverly Teicher lists 54 human sarcoma lines, Twenty six of these are osteosarcomas, 4 are chondrosarcomas and of the remaining 24 cell lines, 12 are paediatric rhabdomyosarcomas. All listed leiomyosarcomas are gynaecological in origin. (Teicher 1999). The 8 adult soft tissue sarcoma cell lines listed have not all been fully characterised and of these lines, the American Type Culture Collection (ATCC) only has available for purchase, the high grade fibrosarcoma cell line HT1080, the low grade fibrosarcoma line, SW684, the liposarcoma line SW872, the synovial sarcoma line SW982 and GCT, the cell line derived from a lung metastasis that are reasonably characterised. There is, however, no cytogenetic data on GCT and the grade of original tumour is not known for SW982 and SW872. One reason for this relative paucity of adult STS cell lines may be senescence. When a tumour specimen is first placed in culture, it is known as a primary culture. Once it is subcultured, it is known as a cell line (Freshney 2000). Senescence is the dominant trait and as such, most cell lines are likely to die out. Alternatively, a cell line may pass through crisis (Reddel 2000) and ‘transform’ to become a continuous immortal cell line. Immortalisation may occur as a result of mutations that occur in vitro, or be an inherent

~~226~~

~~Chapter 5: Characterisation of Cell Lines~~

nature31 of a subpopulation of the explanted neoplastic cells. Different explanted neoplastic tissue may have varying culture requirements that are difficult to predict, leading to suboptimal culture conditions and cell death. Other reasons may include microbial contamination, cross contamination with other cell lines and overgrowth of ‘normal’ fibroblasts. More recently, there have been reports in the literature of newly established cell lines. These include cell lines developed from liposarcomas (LPS) (Wabitsch, Bruderlein et al. 2000; Nishio, Iwasaki et al. 2003a), synovial sarcomas (SS) (Noguchi, Ueki et al. 1997; Sonobe, Takeuchi et al. 1999; Berardi, Parafioriti et al. 2003), malignant fibrous histiocytomas (MFH) (Nakatani, Marui et al. 2001; Nishio, Iwasaki et al. 2003b), various sarcoma cultures (Hu, Nicolson et al. 2002) and the less common clear cell soft tissue sarcomas (CCSST) (Epstein, Martin et al. 1984; Hiraga, Nojima et al. 1997; Crnalic, Panagopoulos et al. 2002). While cell lines are not available to all researchers through ATCC, it is possible that they may be requested from the individual investigators or laboratories where they were developed. More importantly, they could contribute toward a repository of well characterised STS cell lines.

~~5.1.2 Methods of Primary Culture in Literature~~

A variety of methods have been described in the literature pertaining to the primary culture of sarcomas or other malignant tumours. The initial step common to all is sterile dissection of the malignant tissue to 1mm3 pieces. This may be followed by direct seeding to culture dishes, allowing the tissue to attach on its own or facilitating adherence by scratching the dish (Freshney 2000). This simple method proved successful for the establishment of the widely used high grade fibrosarcoma cell line

31 Immortal cell lines are known to possess some form of telomere maintenance mechanism (TMM) that allows the cell to escape senescence. This involves either the enzyme telomerase or an alternate mechanism for lengthening of telomeres (ALT), concepts which were introduced in Chapter One (Section 1.8.4). This will be considered in greater detail again in the subsequent sections of this chapter dealing with chracterisation of two new cell lines. 227

~~Chapter 5: Characterisation of Cell Lines~~

HT1080 (Rasheed, Nelson-Rees et al. 1974). Noguchi and colleagues more recently also used direct seeding of minced tissue to establish their monophasic SS cell line SN- SY-1 (Noguchi, Ueki et al. 1997). A gastrointestinal stromal tumour (GIST) cell line was also reported to be derived in this fashion (Taguchi, Sonobe et al. 2002), as was a MFH cell line, TNMY1 (Nakatani, Marui et al. 2001). Disaggregation of tissue by enzymatic digestion with trypsin is another method of primary culture. HS-SY-3, another synovial sarcoma cell line characterised in 1994 and uterine myxoid MFH cell lines reported in 2001 were established in this fashion (Sonobe, Takeuchi et al. 1999; Kiyozuka, Nakagawa et al. 2001). Collagenase has also been used for digestion of neoplastic tissue, at varying concentrations (200-2000 IU/ml) and for variable periods (30 min to overnight) (Gunawan, Braun et al. 1998; Freshney 2000; Mairal, Chibon et al. 2000; Hu, Nicolson et al. 2002; Nishio, Iwasaki et al. 2003a; Nishio, Iwasaki et al. 2003b). Xenograft models can also be used to generate cell lines, although the logistics involved in coordinating surgery for the patient and obtaining nude mice of the right age is more complex. A clear cell sarcoma cell line was established by serial transplantation in nude mice (Crnalic, Panagopoulos et al. 2002). The above methods of direct seeding and enzymatic digestion were all assessed and optimised in the present study. Unlike primary culture of epithelial tumours, there is little available literature on enhancing sarcoma cell populations in culture, with the exclusion of normal fibroblasts.

~~5.1.3 Methods of Characterisation of Cell Lines in Literature~~

Once a primary culture has been established and successfully subcultured, the goal is typically to determine the phenotypic and genotypic characteristics of the cell line. Cell lines can be compared to the original tumour source on the basis of its immunostaining profile (Noguchi, Ueki et al. 1997; Gunawan, Braun et al. 1998; Crnalic, Panagopoulos et al. 2002; Nishio, Iwasaki et al. 2003a). For cell lines derived from sarcomas of simple karyotype, the recurrent translocation or mutation characteristic of these tumours can be easily demonstrated, for example, Ewing’s sarcoma has a translocation involving

~~228~~

~~Chapter 5: Characterisation of Cell Lines~~

chromosome 22q1232. This provides a definitive means of determining whether the cell line carries the same mutation as the original tumour. Further means of characterisation include defining the growth and invasion properties of the cell line, such as anchorage independent growth (soft agar colony forming assays), and tumorigenicity. The expression of relevant biomarkers has also been evaluated at protein and transcript levels. In addition to conventional karyotyping, chromosomal analyses includes newer techniques such as comparative genomic hybridisation (CGH) (Mairal, Chibon et al. 2000; Nishio, Iwasaki et al. 2003b) to detect chromosomal copy number changes, whole genome chromosome painting (Bayani, Zielenska et al. 2003) or array CGH. Microarray based gene expression profiling has recently been used to re-classify GG-62, a cell line previously thought to be derived from a Ewing’s tumour, as a clear cell sarcoma (malignant melanoma soft parts) cell line (Schaefer, Wai et al. 2002). This reclassification was based on differential gene expression patterns between GG-62 and 6 other Ewing’s sarcoma lines.

~~5.1.4 Specific Aims~~

This chapter describes the establishment of cell lines from fresh soft tissue sarcomas collected over the period from October 2003 to March 2005. The results of a number of short term cultures are presented. , Morphologic characteristics of these cultures on phase contrast microscopy is presented. Further characterisation of two cell lines that survived repeated subculturing for over a year, are presented in detail in Chapters 6 and 7, using the characterisation methods described in Section 5.2 below.

32 The translocations associated with sarcomas of simple karyotype are summarised in Chapter one, Section 1.8.3.1, Table 1.6 and Figure 1.3. Sarcomas that fall into this category include the synovial sarcomas, Ewing’s sarcoma / primitive neuroectodermal tumour (PNET), alveolar rhabdomyosarcoma, myxoid liposarcomas and congenital fibrosarcomas, among others. Sarcomas of complex karyotype are not defined by a single chromosomal rearrangement, but tend instead to have multiple karyotypic anomalies. 229

~~Chapter 5: Characterisation of Cell Lines~~

~~5.2 METHODS~~

~~5.2.1 Collection of tissue~~

~~5.2.2 Primary culture and establishment of cell lines~~

Fresh tissue pieces were minced with scissors using aseptic technique. 1 mm3 explants were seeded directly to 100 mm2 tissue culture dishes and also disaggregated with Collagenase I before transferring to culture flasks. They were cultured in Roswell Park Memorial Institute (RPMI)1640 medium supplemented with 10% foetal calf serum (FCS ; Cell Culture Laboratories, Cleveland, OH), penicillin G (100 U/ml), streptomycin (100 U/ml) and gentamicin (GIBCO BRL, Invitrogen, Carlsbad, CA) (Appendix V).

~~5.2.3 Characterisation of cell lines~~

~~5.2.3.1 Morphology~~

~~5.2.3.2 Immunohistochemistry~~

4 well chamber slides at 2.5 – 3 × 105 cells per well as well as plasma-thrombin clots of the cultured cells embedded in paraffin were created for immunostaining. The primary antibodies used are summarised in Table 5.1 and the details are given in Appendix V.

~~5.2.3.3 Cytogenetics~~

Metaphase chromosome spreads of the cell lines were prepared and analysed. Briefly, 100 μl colcemid was added per 10 ml culture and incubated for 25 min. The supernatant was transferred to the harvest centrifuge tube and 1 ml Trypsin-EDTA was added to the flask. Both were incubated until the cells in the flask had detached. The centrifuge tube

~~230~~

~~Chapter 5: Characterisation of Cell Lines~~

contents were then returned to the flask, mixed and the contents transferred to the centrifuge tube to be centrifuged. The supernatant was removed and KCl (0.075M) was added. The tube was incubated again at 37°C for 15 min then centrifuged. The supernatant was removed and Carnoy’s fixative (3:1 methanol:glacial acetic acid) added and left overnight. The fixative was changed three times before dropping onto clean slides. Karyotype analysis of G banding was performed on 20 cells and a composite karyotype constructed according to the guidelines of the International System for Human Cytogenetic Nomenclature (ISCN 1995) (1995).

~~5.2.3.4 Mutational Analysis~~

~~Genomic DNA was extracted from the cells using a commercially available kit (Qiagen) as per manufacturer’s instructions and 1 μg of DNA was used as the template for the PCR reaction.~~

5.2.3.4.1 TP53 mutation analysis for LMS-LFS cell line PCR was used to amplify the region across exons 5 to 6 of the TP53 gene using the forward primer 5'-ccg tct tcc agt tgc ttt at-3' and the reverse primer 5'-tta acc cct cct ccc aga-3'. The cycle parameters are given in Appendix V.

5.2.3.4.2 KIT and PDGFRA mutation analysis for GIST-M cell line Exons 9, 11, 13 and 17 of the KIT gene on chromosome 4q11-q12 were examined for mutations as was exon 18 of PDGFRA on 4q11-q12. The GIST-M cell line at passages 10 and 18 were analysed. The primers for the mutational analyses are given in Table 5.2 and the protocol is detailed in Appendix V.

~~231~~

~~Chapter 5: Characterisation of Cell Lines~~

Table 5.1. Antibodies used in characterisation of cell lines ",2' -"77.# -30!#)'*32'-, .WUSZRS ,2 *038Q/ D-** Q*8SSTR .WU 6S ,2 *038 SSTR .WU. TVR## ,2 *038 SSTR .WU. VTS 6,!-%#,#Q* + 0'"%#Q8 SSSR .WU. SXTR 6,!-%#,# SSSR .WU 6VY <-4 ! 120 Q<#5! 12*#Q>7 SSTR 7'2#,', . &'$2$0-+773,%7#3,7'+# SSSRRR 78S. ,2 *038SSTRR 8,2'Vα8 '%+ *&#+'! *Q2/-3'1Q6 SSVRR *72-)#0 2',8-S 8-U )-Q&*-1203.Q #,+ 0) SSTRR * [[ )-SSURR * SR <-4 ! 120 Q<#5! 12*#Q>7 SSTRR * UV )-SSSWR :'+#,2', )-SSSRR #1+', )-SSSRR !*T )-SSSWR SRR. )-SSXRRR * SSY 73. )-SSZR .S.-*7!*-, *RS-,-!*-, *R8S+--2&31!*#8!2',R#S773,%7#3,7'+Q #.2.& 0+ !-*-%7Q*&-,, +< 2'-, *>,'4#01'27#"'! *!&--*Q75 ,%(3Q-32&7-0# ##3&#.WU ,2' -"7. TVR+ .12-2&#!#,20 * <8 ',"',%0#%'-,-$2&#.0-2#',5&'!& !-00#1.-,"12-5�#2&#+ (-0'27-$+32 2'-,1-!!30',2&# [Y%#,#TSRR ,"* SSY 0#0 '2 ,2' -"'#1Q5&'*#2�#+ ',"#0 0#+-31# ,2' -"'#1T

~~232~~

~~Chapter 5: Characterisation of Cell Lines~~

Table 5.2 Primers used for mutation analysis of GIST-M cell line #,#( +# 6-, 0'+#0 #/3#,!#1 + ["SW′V33**38&8&388&**8&&&*333VU′WW°* (SW′V3&&38&8*8&8&**3888*83**VU′WW°* SS "SW′V**8&8&3&*3*3883&8*3&8&8*VU′WY°* (SW′V8&****3&333*838*3&8**VU′WV°* SU "SW′V3&*&*33&8*83*8&333&**8&VU′WY°* (SW′V38888&&*8&*33&&8*8*&&*3VU′WY°* SY "SW′V3&&3333*3333*3**3**88*VU′WS°* (SW′V3&*8&&8*3&3*88&*8&8&VU′WV°* /# SZ "SW′V*8&3*33&*8&&&&3&83&VU′WU°* (SW′V&8&8&3888&3&3&&&8&&83&8VU′WW°*

~~5.2.3.5 Analysis of Telomere Maintenance Mechanism (TMM)~~

A modified telomere repeat amplification protocol (TRAP) assay (Kim, Piatyszek et al. 1994; Wright, Shay et al. 1995) was used to detect telomerase activity in the cell line and immunostaining for ALT associated promyelocytic leukaemia or PML bodies (APBs) and TRF2 was carried out to assess if the cell line utilised ALT as its TMM (Henson, Hannay et al. 2005). The TRAP assay was modified from that described previously, in that there was no use of radioactive nucleotides or end labelling of primers. The GIST-M cell line at passage 11 and LMS-LFS cell line at passage 30 were analysed.

5.2.3.5.1 TRAP Assay Protein was extracted from cells using ice cold lysis buffer as described by Kim et al in 1994 (Kim, Piatyszek et al. 1994) and stored in 1μg/μl aliquots for the TRAP assay. Primers used for the PCR were as given below and the cycle parameters are given in Appendix 5: Primer M2 (or TS) 5’-AAT CCG TCG AGC AGA GTT-3’ Primer CX 5’- CCC TTA CCC TTA CCC TTA CCC TAA-3’

5.2.3.5.2 ALT associated PML bodies (APB) staining Cells were cultured on chamber slides, washed and fixed before incubating with anti- PML rabbit antibody (Chemicon, Temecula, CA), followed by anti-rabbit FITC goat

~~233~~

~~Chapter 5: Characterisation of Cell Lines~~

antibody (Sigma, St Louis, MO). Hybridisation with the Cy3 labeled peptide nucleic acid FISH probe for telomeric DNA (Applied Biosystems, Framingham, MA) was then carried out for three hours before counterstaining nuclei with 4′6-Diamidino-2- phenylindole (DAPI) (Sigma). Colocalisation of PML bodies and telomeric DNA within the nuclei of the cells represents presence of APBs.

~~5.2.3.6 Expression of Biomarkers on Real time RTPCR~~

5.2.3.6.1 RNA extraction and cDNA synthesis RNA was isolated using TRIzol® (Invitrogen, Life Technologies. Carlsbad, CA) (Appendix I). The quality and concentration of the RNA was confirmed using both the Agilent 2100 BioAnalyzer (Agilent Technologies, Palo Alto, CA) and the NanoDrop ND-1000 (NanoDrop Technologies, Delaware, USA). cDNA was synthesised using a first strand cDNA synthesis system for RT-PCR (Marligen Biosciences, Ijamsville, MD) according to the manufacturers instructions, using 1 μg of RNA per 20 μl reaction.

5.2.3.6.2 Transcripts selected for analysis The two cell lines fully characterised in this thesis were derived from a gastrointestinal stromal tumour (GIST) and a leiomysosarcoma (LMS). Some of the biomarkers chosen for study were therefore based on transcripts or proteins published in the literature as being characteristic of GISTs. These included protein kinase C theta (PRKCQ), tumour necrosis factor receptor superfamily 6B (TNFRSF6B) and G-protein coupled receptor 20 (GPR20), which has been reported as being differentially expressed in GISTs in a microarray study comparing GISTs and spindle cell tumours (Allander, Nupponen et al. 2001). Phosphodiesterase 2A (PDE2A), known to be responsive to treatment with the KIT tyrosine kinase inhibitor Imatinib, was also chosen as a marker for study (Frolov, Chahwan et al. 2003). A chromatin binding protein, high mobility group box 1 (HMGB1), has also been reported as being expressed in GISTs with KIT mutations and was therefore included in this study (Choi, Kim et al. 2003). Primers were also designed for angiogenic factors such as platelet derived growth factors A and B (PDGFRA and PDGFRB) and vascular endothelial growth factor receptor 2 (VEGFR2/KDR). PDGFRA, KDR and KIT are all located on the same chromosome region (4q11-q12), in addition to which, KDR, PDGFRA and PDGFRB are all, like KIT, tyrosine kinase receptors.

~~234~~

~~Chapter 5: Characterisation of Cell Lines~~

5.2.3.6.3 Primer design Primers were designed using Primer3 software developed by S. Rozen and H. J. Skaletsky (http://www-genome.wi.mit.edu/genome_software/other/primer3.html). Primers were obtained from Sigma Genosys, Castle Hill, Australia. The primer pairs used are shown in Table 5.3. The target genes were quantified by real time PCR (ABI PRISM® 7700 Sequence Detection System, Applied Biosystems) using Platinum® SYBR® Green qPCR SuperMix-UDG (2X) as per manufacturer’s instructions.

5.2.3.6.4 Real time RT-PCR protocol and relative quantitation Real time RT-PCR cycling parameters were as described in Chapter 3 (Section 3.2.4). Relative quantification was carried out as described by Pfaffl in 2001 (Pfaffl 2001), using reference genes for normalisation and the normal fibroblast cell line MRC5 as the control, as described previously in Chapter 3 (Section 3.2.6). The expression ratios were also calculated using the software tool (REST 2005©). The log2 values of these ratios were then analysed using ANOVA to evaluate statistical differences in expression.

~~235~~

~~Chapter 5: Characterisation of Cell Lines~~

~~Table 5.3 Primers used in real time RT-PCR for characterisation of cell lines~~

~~Gene Name Genbank ID Primer Sequences Tm Amplicon Length~~

~~PDE2A NM_002599 F: 5′-TACTCTGCACCTCTGACCACAC-3′ 64.3°C 239 R: 5′-GCTGTTTCTAGTCCTCCCACAG-3′ 63.1°C~~

~~PRKCQ NM_006257 F: 5′-GGAGGGGACTTAATGTACCACA-3′ 63.7°C 216 R: 5′-GGCATCTCCTAACATGTTCTCC-3′ 63.6°C~~

~~PDGFRA NM_006206 F: 5′-AAACAGGAGGAGACTGCAAGAG-3′ 63.7°C 154 R: 5′-GAAGACAGCCTAAGACCAGGAA-3′ 63.4°C~~

~~PDGFRB NM_002609 F: 5′-CCAGCAAGTCTCAAGAACACAG-3′ 63.9°C 247 R: 5′-TGTCCTCACTGTCCATTCTGTC-3′ 64.1°C~~

MAP4K4 NM_004834 F: 5′-CCCTCCTTCCTGTTCCTCTTAT-3′ 63.4°C 215 R: 5′-AGACACCCACAAGACAGAGTCA-3′ 63.8°C ACVR2B NM_001106 F: 5′-CAGCTGGATGTGTAGGTGAAAG-3′ 63.5°C 178 R: 5′-CTCCAGTTCAGAGTCCCATTTC-3′ 63.8°C

~~GPR20 NM_005293 F: 5′-GGTGCTCATCATCTTTCTCGTC-3′ 63.0°C 150 R: 5′-CACTGGTGACGAAGCAGTAGAC-3′ 64.1°C~~

~~TNFRSF6B NM_032945 F: 5′-ACGCTGGTTTCTGCTTGG-3′ 64.1°C 211 R: 5′-AGGGTGTCATGGGAGGAAG-3′ 64.0°C~~

~~HMGB1 NM_002128 F: 5′-AGTGCTCAGAGAGGTGGAAGAC-3′ 63.8°C 193 R: 5′-TCAGAGCAGAAGAGGAAGAAGG-3′ 63.9°C~~

~~KDR NM_002253 F: 5′-GGCCCCTGATTATACTACACCA-3′ 63.5°C 211 R: 5′-GGTAGGCAGAGAGAGTCCAGAA-3′ 63.7°C~~

~~ACTB BC014861 F: 5′-CACCACACCTTCTACAATGAGC-3′ 63.4°C 163 R: 5′-ATAGCACAGCCTGGATAGCAAC-3′ 64.2°C~~

GAPDH NM_002046 F: 5′-GGTGGTCTCCTCTGACTTCAAC-3′ 63.9°C 212 (SW′V*3*33**3*33&3&*3*33&*3VU′ XUT[°* "S"-05 0".0'+#0Q(S(#4#01#.0'+#0Q #XVS&V.0-2#',!-3.*#"0#!#.2-0TRQ WS%'%&+- '*'27%0-3. -6 SQ #S: 1!3* 0#,"-2&#*' *%0-52&$ !2-00#!#.2-0T:-&"(TQ X S .&-1.&-"'#12#0 1# T8Q #$@S .0-2#', )', 1# * 2 Q /# ," /#S .* 2#*#2 "#0'4#" %0-52& $ !2-0 0#!#.2-01 8 ," Q 2/#/\S 33+-30,#!0-1'1$ !2-00#!#.2-013.#0$ +'*7XQ S&*7!#0 *"#&7"#VUV.&-1.& 2#"#&7"0-%#, 1#T&#,# 17+ -*131#" 0#2&-1# ..0-4#" 72&#%3+ ,&#,-+#60% ,'1 2'-,&#,#<-+#,!* 230#*-++'22##%&<*T

~~5.3 RESULTS~~

~~5.3.1 Establishment of Primary cultures and Short term cell lines~~

The tumours from which the primary cultures were seeded over the period from October 2003 to March 2005 are summarised in Table 5.4. The cultures were assigned an identifying numbered code in chronologic order, with a “DOS” placed as a prefix to designate the department of surgery. Cell lines that went on to be fully characterised

~~236~~

~~Chapter 5: Characterisation of Cell Lines~~

were then given new names. Clinical details of the patients and the source tumours (DOS 5 to DOS25) are provided in Appendix V. As noted in Table 5.4, DOS15, 16, 20, 21, 23, 24 and 25 still remained in culture at the time of writing. DOS13 was frozen down at passage 14 and not characterised further due to time limitations. DOS5 to DOS12 and DOS14 senesced at early passages, with insufficient cell numbers to determine proliferation rate. Some of the others that remained in culture at the time of writing still had mixed populations of cells that would require selective trypsinisation for enrichment of tumour cell population and as such, the doubling time of these cells was yet to be determined.

~~5.3.2 Morphology of Short Term Cultures~~

~~5.3.2.1 Liposarcoma (LPS) cultures~~

In general, the liposarcomas, well differentiated as well as the myxoid variants (DOS5, 6, 7 and 19) tended to be very slow growing. The source tumours were hypocellular with infrequent malignant lipoblasts. As a consequence, the explants largely comprised fat globules and few cells. The cells in culture did initially resemble lipoblasts, with prominent nuclei and nucleoli and cytoplasmic vacuoles containing fat droplets. DOS19, however, was received in culture from Canberra after one month in culture and appeared to resemble normal fibroblasts only. These short term cultures did not survive subculturing and were thus not characterised further. The exception was DOS13 (Figure 5.1 (e)), where collagenase digestion of tissue fragments from a low grade myxoid LPS yielded a dense population of cells that were rapidly subcultured to passage 15 before cryopreservation.

~~5.3.2.2 Malignant fibrous histiocytoma (MFH) and Pleomorphic sarcoma (PMS) cultures~~

The changing concept of MFH was introduced in Chapter 1 (see Section 1.4.2). Current theory holds that MFHs represent part of a continuum of fibroblastic tumours through to its most undifferentiated pleomorphic variant. As such, the cultures initiated from tumours described as MFH, PMS and undifferentiated sarcomas are considered together. The earlier cultures initiated from high grade undifferentiated or pleomorphic

~~237~~

~~Chapter 5: Characterisation of Cell Lines~~

sarcomas (DOS8 and 9), were minced and seeded directly to culture dishes without collagenase digestion. Table 5.4 Primary Cultures initiated October 2003 - March 2005 2'12-*-%7-$ -30!#3+-30 0 "# -"# 11 %#(3+ #0 /.*'.-+ V*')# 5#**"'$$#0#,2' 2#" S 6W T 76-'"/.S 6XT /.*'.-+ V*')# 5#**"'$$#0#,2' 2#" S 6Y S T .U 6ZS T "% .U 6[ [U V 612#-1 0!-+ U 6SR## <#30-$' 0-1 0!-+ S 6SST >,"'$$#0#,2' 2#"1 0!-+ U 6ST V 76-'"/.S 6SUSV " * 6SVX &3%'%& 6SW#TY /U 6SX#VR -+ 07-, *($ − 6SYV #,'%,1!& #+'!$ 1!''2'1$ − 6SZV 76-'"/.T 6S[##U '.& 1'!U 6TRU -5',%_1 .<-3$−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� 2& , 1-$22'113#1 0!-+ T 6S[ ..# 0#"2--,*7& 4#,-0+ *$' 0- * 121',!3*230#5&'!&1#,#1!#"T$<- &'12-*-%'!%0 "',%5 1 11'%,#"2-2#! 1#1S%'12-*-%'!%0 "',%"-#1,-2 ..*72--5',%_11 0!-+ .<-3 12&#,-0+ *!#**!-3,2#0. 02'13,),-5,R',2&#-2�! 1# 6SYQ2&#. 2'#,2& "0#!#'4#",#- "(34 ,2 !&#+-2� .7T<-%0 "#'1 ..*'#"2- #,'%,*#1'-,1 6SZT

DOS8 demonstrated slow growth, the first subculture being performed after a number of weeks. Normal fibroblasts eventually predominated and the cell strain was not maintained in culture. In contrast, there was rapid outgrowth from the explants in the culture designated DOS9. The cells were spindle shaped to polygonal with short but slender cytoplasmic processes. They grew in irregularly piled foci and had oval nuclei with prominent nucleoli (Figure 5.1 (a)).

~~238~~

~~Chapter 5: Characterisation of Cell Lines~~

~~" # $~~

Figure 5.1 Earlier Primary Cultures as seen on phase contrast microscopy (a) DOS9: Outgrowth of spindle cells from explanted fragment (bottom right corner of image) of a high grade malignant fibrous histiocytoma. (b) DOS10: Cultured cells received from Canberra, directly seeded from an Extraskeletal osteosarcoma. The cells have long processes, small nuclei and prominent nucleoli. (c) DOS11: Tissue obtained from a low grade neurofibrosarcoma. The culture was very slow growing and did not survive repeated passaging. (d) DOS12: Tissue from an undifferentiated high grade sarcoma was obtained and culture initiated by direct seeding as well as digestion with collagenase. The spindle cells were cultured for 17 months before freezing down. (e) DOS13: Collagenase digestion of tissue from a low grade myxoid liposarcoma resulted in a good initial yield of cells that were passaged at least 15 times before freezing down. (f) DOS14: Dense outgrowth from explanted follicular dendritic cell sarcoma involving axillary lymph nodes. The cells were maintained in continuous culture and passaged 7 times before freezing down.

Subculturing was carried out successfully. Tissue fragments from a high grade undifferentiated sarcoma were obtained. Outgrowths from explants seeded directly to dishes were also cultured (DOS12) (Figure 5.1(d)). These cells were also spindle shaped to polygonal in morphology, growing in whorls and fascicles. The nuclei were ovoid, with distinct nucleoli.

~~239~~

~~Chapter 5: Characterisation of Cell Lines~~

In contrast to DOS12, the next cell culture derived from a grade 3, undifferentiated sarcoma (DOS23) did yield a dense population of rapidly proliferating cells following collagenase digestion. The spindle shaped cells with slender cytoplasmic processes were arranged in fascicles, resulting in a storiform pattern (Figure 5.2).

Figure 5.2 Morphology of cell line derived from an undifferentiated sarcoma (DOS23) (a) shows the spindle shaped cells with long processes as seen on phase contrast microscopy. (b) The cells at passage 15 were cultured on chamber slides, allowed to adhere over 48 hours, then washed and fixed before staining with haematoxylin and eosin (H&E). The same spindle cell morphology is seen, with the cells growing in interlacing short fascicles (storiform pattern). The nuclei are large and nucleoli are prominent.

The last pleomorphic sarcoma to be placed into culture, DOS24, was the only cell strain derived from a metastatic tumour. The patient had received chemoradiotherapy following resection of previous lesions but the present lesion had arisen outside the irradiated field. Collagenase digestion again yielded large cell numbers. In this case, early passages comprised spindle cells and pleomorphic cells, as well as the occasional multinucleate giant cell (Figure 5.4 (a) and (b)). The giant cells gradually disappeared and the spindle cells became dominant.

~~5.3.2.3 Extraskeletal Osteosarcoma and Neurofibrosarcoma Cell Cultures~~

DOS10 was received in culture from our collaborators in Canberra while the histopathology of the source tumour was still being determined. In culture, the cells had polygonal cell bodies and numerous long slender cytoplasmic processes. While the cytoplasmic membrane was somewhat indistinct, the nuclei and nucleoli were

~~240~~

~~Chapter 5: Characterisation of Cell Lines~~

prominent. This cell strain was frozen down once the tumour of origin was confirmed to be an osteosarcoma. DOS11 was placed into culture by directly seeding tissue obtained from an excision biopsy of a lower limb lesion that was initially reported as a neurofibroma and later revised to a low grade neurofibrosarcoma. The spindle cells with multiple cytoplasmic processes cells proliferated slowly and senesced following the first subculture.

~~5.3.2.4 Follicular Dendritic Cell Sarcoma (FDCS) Cell Culture~~

The cultured cells from this rare metastatic FDCS deposit proliferated slowly, forming interlacing fascicles. The cells were spindle shaped, with indistinct cell borders but distinct ovoid nuclei and prominent nucleoli. The culture did not, however, successfully pass through crisis and become immortal.

~~5.3.2.5 Embryonal Rhabdomyosarcoma Cell Culture~~

The cultured cells in this case were derived from a rhabdomyosarcoma that had been subjected to neoadjuvant chemoradiation. Although the histopathologic report confirmed the presence of viable tumour cells, the cell strain was comprised predominantly of fibroblasts. Embryonal rhabdomyosarcomas fall into the group known as the small round blue cell tumours but no small round cells were seen in culture. No spindled rhabdomyoblasts were discernable.

~~5.3.2.6 Biphasic Synovial Sarcoma (SS)~~

Cells from a biphasic synovial sarcoma demonstrated the biphasic characteristics of the source tumour, in that there were populations of cuboidal epithelial cells growing in solid nests (Figure 5.3(a)) intermixed with plump spindle cells, forming fascicles, such as those shown in Figure 5.3(b). The epithelial cells possessed large oval nuclei and were occasionally binucleate. Mitotic figures were also observed in both cellular components.

~~241~~

~~Chapter 5: Characterisation of Cell Lines~~

~~! "~~

Figure 5.3 Synovial sarcoma and PNET Cultures Two of the primary cultures initiated in 2005 as seen on phase contrast microscopy are depicted above. Both direct seeding to tissue culture dishes and enzymatic digestion with collagenase was carried out. The top panel (a) and (b) were initiated from the same biphasic synovial sarcoma (DOS20), with (a) showing outgrowth from a directly seeded explant and (b) showing the cell suspension seeded after collagenase digestion. The outgrowth in (a) had closely packed nests of plump cells with large nuclei and granular cytoplasm, whereas collagenase digestion yielded a suspension that was then dispersed evenly. Once cell proliferation had resulted in greater cell density, the cells had a similar plump appearance. The bottom panel shows (c) outgrowth from directly seeded explants and (d) cells placed in culture after collagenase digestion of the same Ewing’s / primitive neuroectodermal tumour (DOS21) . The latter were more spindle shaped, resembling fibroblasts and were subsequently lost as a result of microbial contamination. The subcultures from (c) are still maintained at the time of writing. These cells were small, with little cytoplasm and oval nuclei.

~~242~~

~~Chapter 5: Characterisation of Cell Lines~~

~~! "~~

Figure 5.4 Primary cultures after collagenase digestion The top panel demonstrates the heterogeneous cell populations cultured from the same metastatic pleomorphic sarcoma (DOS24). (a) Polygonal malignant cells with granular cytoplasm, large nuclei and prominent nucleoli, tending to grow as nests are seen. Interspersed are occasional multinucleate giant cells. (b) Spindle shaped cells growing in fascicles within the same tissue culture flask, resembling fibroblasts. The plump cells eventually apoptosed, leaving the spindle shaped cells to proliferate. The bottom panel shows cells in culture from the same low grade leiomyosarcoma (DOS25). (c) These cells were seeded after 2 hours of collagenase digestion, once a cloudy cell suspension was seen. (d) These slightly more plump cells were present in a separate flask seeded with a cell suspension after overnight collagenase digestion.

~~243~~

~~Chapter 5: Characterisation of Cell Lines~~

~~5.3.2.7 Extraskeletal Ewing’s / Peripheral Neuroectodermal Tumour (PNET) cell culture~~

Early passages of this cell strain comprised both malignant cells and fibroblasts, but after selective trypsinisation, the malignant cells predominated. The malignant cells were uniformly small, round and densely packed (Figure 5.3 (c) and (d)). The oval nuclei occupied much of the cell body. Numerous mitotic figures were noted. The cell strain had a rapid proliferation rate and appeared to pass through crisis within the first few cultures, with large numbers of non viable floating cells between adherent nests of cells. This cell line remains in culture at the time of writing and warrants further characterisation. Karyotyping and immunohistochemical analysis are ongoing.

~~5.3.2.8 Leiomyosarcoma cell culture~~

Cells were also cultured from a sample of a low grade leiomyosarcoma. The cultured cells, DOS25, proliferated slowly and at the time of writing, had only been subcultured a few times. The cells were spindle shaped with slender cytoplasmic processes, ovoid nuclei and distinct nucleoli. The cells were arranged in short interlacing fascicles (Figure 5.4 (d) and (d)).

~~5.4 DISCUSSION~~

A number of short term cultures were initiated, from a diverse range of soft tissue sarcomas. Different techniques were trialled, from direct seeding of minced tissue, to enzymatic digestion with trypsin or collagenase. In general, the cells proved to be too sensitive to trypsin and the initial yield of cells from this technique was low. It was therefore not used for the later cultures. Direct seeding had greater success, typically with good outgrowths from explanted tissue. The rate of growth, however, was slow with this method and the first subculture usually necessitated long periods of trypsinisation to lift adherent cells from the culture flasks. This consequently resulted in loss of some cells. Collagenase digestion usually resulted in the greatest number of cells in the cloudy suspension, which could then be gently centrifuged, resuspended in culture medium and seeded to flasks. Subsequent subculturing with trypsinisation was also easier with this method.

~~244~~

~~Chapter 5: Characterisation of Cell Lines~~

~~5.4.1 Problems with maintaining cultures~~

~~5.4.1.1 Bacterial and Fungal contamination~~

The rate of bacterial contamination was slightly higher with the collagenase digestion technique, as more handling of the tissue and preparation of collagenase was required. For the cultures DOS20 and 21, the collagenase digested flasks were lost to contamination, leaving the flasks containing directly seeded tissue to be subcultured. To reduce the contamination rate, the initial culture medium included penicillin/streptomycin, gentamicin and fungizone.

~~5.4.1.2 Overgrowth of normal fibroblasts~~

Cultures where normal fibroblasts appeared to predominate were frequently encountered. There is no guaranteed method of excluding these from culture, although some success was achieved with quick trypsinisation. The normal fibroblasts were more adherent, while the malignant cells lifted more easily and quickly. This enabled the malignant cells to be harvested, leaving fibroblasts attached to the original flask. Enriched malignant cell populations could thus be obtained with each quick trypsinisation, ultimately resulting in a cell line comprised solely of malignant cells. DOS15, later renamed GIST-M, DOS16, renamed LMS-LFS and DOS21 were successfully cultured in this manner.

~~In some cases initial collagenase digestion resulted in greater numbers of normal fibroblasts being cultured, while direct seeding yielded higher numbers of malignant cells (DOS20).~~

~~5.4.1.3 Crisis and Senescence~~

The major barrier to the development of continuous cell lines is senescence. The majority of cultures will eventually senesce. Senescence is thought to occur after 30-60 population doublings (Reddel 2000). While cell lines may be immortalised by transfecting them with DNA viruses such as SV40, human papilloma virus, Epstein- Barr virus or the fungal metabolite aflatoxin (Tsutsui, Fujino et al. 1995; Rhim 2000; Tsutsui, Kumakura et al. 2002; Tsutsui, Kumakura et al. 2003), transfection in itself may alter the gene expression profile of the cells and disrupt many cellular pathways.

~~245~~

~~Chapter 5: Characterisation of Cell Lines~~

A small proportion of cultures reach crisis and spontaneously activate mechanisms that enable them to escape senescence. As discussed in Chapter 1 (Section 1.8.3), this is thought to involve, in part, telomere maintenance mechanisms (TMMs). From the series of cultures established in this study, TMMs were examined in two of the cultures that survived repeated subculturing and continued to proliferate twelve months after being initiated. These will be discussed further in the subsequent sections of this chapter.

~~5.4.2 Selection of cell lines to further characterise~~

In addition to the two cultures (GIST-M and LMS-LFS) to be examined in greater detail in the subsequent chapters, a number of other cell strains, namely, DOS 13, 20, 21 and 23 also remain in culture at the time of writing. As these have been subcultured many times and have been in culture for six months to a year, they may potentially have passed through or be nearing crisis. Traditionally, whether or not a cell line will senesce cannot be truly determined unless more than 50 doublings have occurred.

In the case of the above cell lines, TMMs may be evaluated to address this question. It is possible, however, that telomerase and/or ALT may not be activated or expressed until crisis is reached. In addition, cytogenetics may provide evidence of malignant cell predominance as normal fibroblasts are unlikely to have multiple clonal anomalies. Once these features are established, further characterisation of the cell lines can be carried out. However, these cell cultures may still function as useful in vitro models, as demonstrated by the use of the short term cultures GIST544 and GIST780 to study the mechanisms of KIT activation and the effects of imatinib treatment (Tuveson, Willis et al. 2001; Duensing, Medeiros et al. 2004). Consistent amounts of early passages of cell lines can be cultured for use. Two of the cell lines established in this study, DOS15 or GIST-M and DOS16 or LMS-LFS did continue to proliferate after 18 months in culture. These were characterised more fully and are therefore considered in greater detail in the following two chapters.

~~246~~

~~Chapter 6: GIST-M Cell Line~~

~~2" 9~~

~~6. Characterisation of a novel GIST-M cell line from a Gastrointestinal Stromal Tumour (GIST) arising from the Duodenum~~

~~247~~

~~Chapter 6: GIST-M Cell Line~~

~~6.1 INTRODUCTION~~

In the previous chapter, the morphology of some newly established sarcoma cell strains were described. This chapter deals with the complete characterisation of a GIST cell line derived from a tumour arising from the duodenum. The methods employed in this process are detailed in Chapter 5, Section 5.2 and Appendix V. Prior to presenting the results, some of the key concepts in GIST pathology are introduced below.

~~6.1.1 Clinical Presentation~~

GISTs are the most common mesenchymal tumour of the gastrointestinal (GI) tract. As indicated in Chapter 1 (Section 1.4.2), the usual histologic grading system does not apply for GISTs. There is a spectrum of tumours from benign to malignant, with “malignant” GISTs being diagnosed on the basis of tumour size (>5 cm) and mitotic rate (>2−5/10HPF). Tumour size, location, mitotic rate and rupture are associated with poor prognosis. Surgical resection is the treatment of choice and complete resection is associated with a 5 year actuarial survival rate of 54% (DeMatteo 2002; DeMatteo, Heinrich et al. 2002). For a tumour larger than 10 cm, the 5 year survival is less than 20% (DeMatteo, Lewis et al. 2000). Standard chemotherapy for recurrent or disseminated disease with combination of agents such as doxorubicin, dacarbazine, mitomycin and cisplatin has poor results (DeMatteo, Heinrich et al. 2002; Demetri 2002), while intraperitoneal chemotherapy with mitoxantrone has had limited success in a small series of patients (DeMatteo 2002).

~~6.1.2 Histogenesis~~

Previously grouped together with leiomyosarcomas, GISTs were thought to have a common histogenesis from GI smooth muscle cells. More recently, GISTs have been recognised as a distinct entity, sharing genetic and phenotypic characteristics with the pacemaker cell of the gut, known as the interstitial cell of cajal (ICC) (Kindblom, Remotti et al. 1998). The majority of GISTs stain positively for KIT (CD117), a type III transmembrane tyrosine kinase receptor, known to be expressed by ICCs. These pacemaker cells are thought to arise from pluripotent mesenchymal cells, which are capable of differentiating either into ICCs or smooth muscle cells. KIT expression is a

~~248~~

~~Chapter 6: GIST-M Cell Line~~

~~prerequisite for differentiation to ICCs and it was originally also claimed as a prerequisite for diagnosing GISTs (Fletcher, Berman et al. 2002).~~

~~6.1.3 KIT Mutations~~

The KIT gene encodes a transmembrane tyrosine kinase receptor, a member of a family of receptor tyrosine kinases (RTKs) that includes PDGFRA, PDGFRB, FLT3 and vascular endothelial growth factor receptor 2 (VEGFR2). The receptor possesses an extracellular ligand binding domain, a juxtamembrane domain, and an intracellular domain that includes the split tyrosine kinase residues and a kinase insert. In non neoplastic tissue, its ligand is stem cell factor (SCF). Ligand binding causes dimerisation and activation by autophosphorylation at the tyrosine residues (Figure 6.11 (a)). Pathways activated in normal KIT function include the PI3K/AKT cascade involved in anti-apoptosis and chemotaxis, the Sos/Ras/ERK cascade involved in cell proliferation and the Jak/STAT pathway (Heinrich, Rubin et al. 2002; Dibb, Dilworth et al. 2004).

In 1998 mutations in the KIT gene were identified in 5 GISTs, all in the juxtamembrane encoding region (exon 11) of the gene (Hirota, Isozaki et al. 1998). Since then, more mutations within exon 11 and those in other exons (9, 13 and 17) of the gene have been described (Lasota, Jasinski et al. 1999; Taniguchi, Nishida et al. 1999; Lux, Rubin et al. 2000). Exon 9 encodes an extracellular domain, while exons 13 and 17 encode the intracellular split tyrosine kinase domains. A minority (6-35%) of GISTs with no demonstrable KIT mutation are known to manifest mutations in the encoding region for platelet derived growth factor A (PDGFRA), also a receptor tyrosine kinase (RTK) (Heinrich, Corless et al. 2003b; Debiec-Rychter, Dumez et al. 2004). Of all these, exon 11 mutations remain the most frequent, with an incidence of up to 92% (Heinrich, Corless et al. 2003a; Tornillo and Terracciano 2006). The majority of GISTs are sporadic and occur between the ages of 40 and 80 years (DeMatteo 2002), but familial GISTs involving germline KIT mutations have also been described (Nishida, Hirota et al. 1998; Isozaki, Terris et al. 2000; Beghini, Tibiletti et al. 2001). In both cases, mutations in the KIT gene result in constitutive activation of the transmembrane receptor KIT (Heinrich, Rubin et al. 2002).

~~249~~

~~Chapter 6: GIST-M Cell Line~~

~~6.1.4 Clinical Relevance of KIT Mutations in GISTs~~

The identification of gain-of-function KIT mutations not only revolutionised the treatment of GISTs, it firmly focused research in GISTs and other sarcomas on the development of targeted therapies33 . Imatinib mesylate (STI 571/Gleevec®/Glivec®, Novartis, Basel, Switzerland), a tyrosine kinase inhibitor, has in recent years been used in the treatment of GISTSs, with partial response rates as high as 80-90% (van Oosterom, Judson et al. 2001; DeMatteo 2002; van Oosterom, Judson et al. 2002; Heinrich, Corless et al. 2003a). This prompted further research into the mechanisms of action of KIT and associated molecular markers. Gene expression profiling has been one of the techniques employed in the discovery of the molecular signature of GISTs (Allander, Nupponen et al. 2001; Nielsen, West et al. 2002; Antonescu, Viale et al. 2004)34. Others have concentrated on determining the phenotypic and prognostic significance of the various KIT mutations (Antonescu, Viale et al. 2004; Penzel, Aulmann et al. 2005). Examining the mechanisms of action of KIT and Imatinib has, however, required the use of cells lines (Tuveson, Willis et al. 2001; Frolov, Chahwan et al. 2003; Duensing, Medeiros et al. 2004).

~~6.1.5 Cytogenetic aberrations and GISTs~~

Several cytogenetic anomalies were found in a study of 19 GISTs to be associated with the “high risk” or recurrent GISTs (Gunawan, Bergmann et al. 2002), suggesting that additional anomalies are acquired as the tumours progress. Similar findings were noted

33 The molecular biology of soft tissue sarcomas in relation to targeted therapy was introduced in Chapter One (Section 1.9) 34 Gene expression arrays identifying gene clusters overexpressed in GISTs were discussed in Chapter One (Section 1.10.3.2). The expression of some of these markers, including protein kinase C theta (PRKCQ), tumour necrosis factor superfamily 6B (TNFRSF6B) and phosphodiesterase 2A (PDE2A) was examined by real time RT-PCR as part of the characterisation of the cell lines GIST-M and LMS-LFS. This is presented in greater detail in the preceding section on methods (Section 5.2) 250

~~Chapter 6: GIST-M Cell Line~~

in two separate studies employing comparative genomic hybridisation to evaluate DNA copy number changes (El-Rifai, Sarlomo-Rikala et al. 2000; Langer, Gunawan et al. 2003). It appears clear that KIT mutations are integral to the ongogenic process in the majority of GISTs. The key events involved in the progression of this disease, are less clearly defined and further in vitro studies are needed to evaluate the pathogenesis of GIST progression and recurrence.

~~6.1.6 GIST cell lines~~

As noted earlier (Section 5.1.1), there is a paucity of sarcoma cell lines available for in vitro studies and there are even fewer reports of characterised GIST cell lines. The first report of a GIST cell line appears in a study examining the in vitro effects of imatinib treatment (Tuveson, Willis et al. 2001). The cell line, designated GIST882 was derived from a metastatic GIST and known to harbour a less common exon 13 KIT missense mutation (K642E). The focus of this previous study was to establish the anti- proliferative effects of imatinib on GIST882 compared to other sarcoma cell lines. Another cell line GIST780, which did not survive continuous passaging, was also briefly mentioned. GIST882 has subsequently been used by other researchers to elucidate the mechanisms of KIT signal transduction in GISTs (Duensing, Medeiros et al. 2004) and investigate changes in gene expression profiles with in vitro imatinib treatment (Frolov, Chahwan et al. 2003). Duensing and colleagues, who focused on the STAT3, AKT and p42/44 MAPK downstream pathways which are thought to be activated by KIT, include in their experiments a short term culture named GIST544. . A cell culture named GIST544 was reported by these authors but the period of time that it was able to be maintained in culture, the source tumour from which it was derived, or its other immunophenotypic characteristics were not specified. GIST554 was described only as a stem cell factor (SCF) dependant culture exhibiting an exon 9 (503_504insAY) mutation. There is only one report of a well characterised GIST cell line (GIST-T1) (Taguchi, Sonobe et al. 2002). Morphology, tumorigenicity, immunostaining profile and karyotype were described for the cell line, which was derived from a pleural metastasis. A brief mention was made of the preliminary KIT mutation analysis, revealing a

~~251~~

~~Chapter 6: GIST-M Cell Line~~

deletion of 57 bases in exon 11. With the emergence of acquired resistance to imatinib treatment (Desai, Maki et al. 2004; Antonescu, Besmer et al. 2005; De Giorgi and Verweij 2005; Debiec-Rychter, Cools et al. 2005), the need for further research in this area and the usefulness of in vitro models has been highlighted. The prognostic significance of the various KIT mutations need to be better defined and secondary acquired mutations investigated. Certainly the concept of acquired mutations in the progression of malignant disease in general is well known and has been discussed in Chapter 1 (Section 1.10.2).

~~6.2 RESULTS~~

This section describes the morphologic, immunophenotypic, karyotypic and molecular characterisation of a GIST cell line derived from a retroperitoneal tumour, resected following six months of neoadjuvant treatment with imatinib. The expression of a number of KIT associated biomarkers were examined using real time RT-PCR. Telomerase and alternate lengthening of telomere (ALT) activity was also investigated. The methods have been described previously in Chapter 5, Section 5.2 and Appendix V.

~~6.2.1 Tumour Source~~

~~The clinical details of the 62 year old patient and the histology of the tumour are presented in Appendix V.~~

~~6.2.2 Establishment of Cell Line~~

Tumour fragments were digested in collagenase overnight, centrifuged and resuspended in media. The initial yield of cells was high, with malignant cells forming distinct colonies or nests. After the first subculture, selective trypsinisation was performed, as described in Appendix V. The more adherent fibroblasts were left on the original flask. This enriched the malignant cell population in the next subculture.

~~6.2.3 Morphology~~

The cell line displayed predominantly spindle cell morphology, with cells arranged in short interlacing fascicles or whorls. The nuclei were uniformly ovoid, with visible nucleoli. Earlier passages of the cell line had mixed populations of spindle cells as well as plumper, epithelioid cells forming nests, as shown in Figure 6.1 (b).

~~252~~

~~Chapter 6: GIST-M Cell Line~~

~~! "~~

Figure 6.1 Phase contrast microscopy and H&E staining of GIST-M cell line (a) Cells after first subculture following collagenase digestion of original tumour and quick trypsinisation to obtain enriched tumour cell population. The occasional fibroblast is still seen. At passage 3, both spindle cells and more rounded epithelioid cells (b) are seen. The cells at passage 6, which were used to perform cytogenetic analysis, are shown in (c). The cells at passage 24, where the proliferation rate had slowed, are shown in (d). The cells were cultured on chamber slides cultures for H&E staining. The image below (e), shows the characteristic morphology of this cell line on H&E staining. The cells are spindle shaped, growing in dense nests and interlacing sheets or fascicles (herringbone pattern), with prominent nuclei and visible nucleoli. #

~~253~~

~~Chapter 6: GIST-M Cell Line~~

~~! "~~

Figure 6.2 Immunocytochemistry for GIST-M cell line The Gastrointestinal stromal tumour (GIST) cell line was cultured on chamber slides in order to perform immunostaining. The cell line was examined for the same markers as the original tumour from which it was derived, to determine if it retained the same phenotypic characteristics. Some of these are shown in the panel above, namely very weak staining for cytokeratin AE1/AE3 (a), weak positive staining for desmin (b), and negative staining for both S100 (c) and smooth muscle actin (SMA) (d). The expression of KIT (CD117), Vimentin and CD34 were also assessed. KIT and Vimentin expression is shown in the following figure. The immunostaining profile of the cell line was found to correlate with that of the tumour from which it was established. As reviewed by Corless and colleagues, approximately 95% of GISTs stain positively for KIT, while a smaller proportion stain for other markers, such as CD34 (70%), SMA (35%), S100 (10%) and desmin (5%) (Corless, Fletcher et al. 2004). In this case, both the tumour and the cell line were negative for CD34, S100 and SMA and positive for desmin, vimentin and KIT.

~~254~~

~~Chapter 6: GIST-M Cell Line~~

~~! "~~

Figure 6.3 cKIT and Vimentin Expression Immunostaining of Gastrointestinal stromal tumour (GIST) cell line was performed by culturing the cells on chamber slides, then fixing and staining them using the alkaline phosphatase method, with fuschin as the chromogen. Red indicates positive staining and nuclei are counterstained blue with haematoxylin. The top panel (a) and (b) shows positive staining of the cell line at passage 7 for KIT (CD117) and Vimentin respectively. The same markers are shown on the bottom panel for the cell line at passage 21 in (c) and (d), where the intensity of staining is considerably weaker. The cell line at this stage had been in continuous culture for a year. The reduced intensity of staining is possibly due to a changing phenotype of the cells in culture, or may be an indication of a mortal cell line nearing senescence. At the time of writing, the cell line remains in culture at passage 28 and has not yet senesced.

~~6.2.4 Immunocytochemistry and Immunohistochemistry~~

The immunophenotypic characteristics of the cultured cells mirrored that of the source tumour, showing weak positive staining for desmin and negative staining for both S100 and smooth muscle actin (SMA) (Figure 6.2. Staining for CD34 was negative, while

~~255~~

~~Chapter 6: GIST-M Cell Line~~

both cell line and source tumour stained positive for KIT and Vimentin (Figure 6.3). Approximately 95% of GISTs stain positively for KIT, while a smaller proportion stain for other markers, such as CD34 (70%), SMA (35%), S100 (10%) and desmin (5%) (Corless, Fletcher et al. 2004). KIT expression was maintained in early passages of the cell line but was negative by passage 21.

~~6.2.5 Cytogenetics~~

17 metaphases from cells at passage 6 were analysed with a banding resolution of 400 bands. A composite karyotype was constructed, as shown in Figure 6.4. Based on ISCN convention, the karyotype can be written as

-2`$$`)*7+b+16+*b6+*`58:).(,b36++_,b5++`&)).6--`0`)*7+*9+* b+/6,,b6,.`)*7+,91b+,5+/b6,.`&))+-5++`8+.`8+/`8+1`8+3`8,+`8 ,, +* 210`.)*2=, (51 = where der refers to derivative chromosomes resulting from translocations t refers to translocations of whole or parts of chromosomes, psudic is pseudodicentric, referring to a chromosome with 2 centromeres (dicentric) where one is suppressed/inactive. The appearance of the chromosome has only one constriction at the active centromere; add describes additional material of unknown origin translocated onto the specified breakpoint idem refers to the second clone having the same abnormalities as the primary (first) clone. Abnormalities restricted to the second clone by convention are described after “idem” and cp identifies the karyotype as a composite, as not all cells are the same.

Two related abnormal clones were detected. The stemline was hypodiploid and contained multiple clonal abnormalities. 38XX implies there is loss of 7 chromosomes, including chromosomes 6, 14, 19, 21 and 22. There was also suspected loss of chromosome regions 1p, 4q33->qter, 9p, 10q22->qter, 15pter->q22, possibly 17p13- >pter, as well as 18q. As there were a number of regions that could not be confidently assigned by G-banding, some of the regions assumed to be lost may be present in these derivative chromosomes. There was apparent gain of 7p15->pter. There were

~~256~~

~~Chapter 6: GIST-M Cell Line~~

translocations involving chromosomes 1 to 17, 2 to 9, 7 to 12 and 10 to 15. The second abnormal clone was a hypotetraploid population resulting from doubling of the stemline with no additional clonal abnormalities observed. Hypodiploidy appears to be a common feature of GISTs. Non-random chromosome abnormalities have been reported in GISTs including loss of chromosome 14 and 22 as primary karyotypic abnormalities. Secondary karyotypic abnormalities described as non-random included loss of 1p, 3p, 9p, 11p, 13q, 10q and/or 15. A combination of at least 3 secondary cytogenetic abnormalities has been shown to characterise high-grade malignant GISTs with aggressive clinical behaviour (El-Rifai, Sarlomo-Rikala et al. 2000; Gunawan, Bergmann et al. 2002; Heinrich, Rubin et al. 2002). This is the pattern that was observed in this GIST-M cell line.

~~257~~

~~Chapter 6: GIST-M Cell Line~~

Figure 6.4 Karyotype of GIST-M cell line Two related abnormal clones were detected. (a) The stemline was hypodiploid and contained multiple clonal abnormalities at the positions indicated by the arrows. (b) The second abnormal clone was a hypotetraploid population resulting from doubling of the stemline with no additional clonal abnormalities observed. Loss of chromosomes 14 and 22 was noted, aberrations known to be associated with GISTs. Secondary abnormalities included loss of regions 1p and 9p, as well as several translocations as detailed in the text.

~~258~~

~~Chapter 6: GIST-M Cell Line~~

~~6.2.6 KIT and PDGFRA Mutation Analysis~~

Exons 9, 11 (codons550-580), 13 and 17 of the KIT sequence were examined, together with exon 18 (codons 816-820) of the PDGFRA sequence in both the cell line at passages 10 and 18 and the original tumour. An insertion mutation was detected in exon 11 (1720_1725 ins GCCCCA). This translates to the insertion of two amino acids, serine and proline (573_574 ins SP) (Figure 6.5). Exon 11 mutations, corresponding to the juxtamembrane region of the KIT receptor, are the most common documented mutations the GISTs. The spectrum of mutations seen in GISTs and the clinical implications of these will be considered further in the discussion.

~~6.2.7 Expression of Biomarkers on Real Time RT-PCR~~

The expression of several biomarkers was assessed by real time RT-PCR for both this cell line GIST-M and the leiomyosarcoma cell line LMS-LFS discussed in the subsequent sections. GIST-M at passage 5 and LMS-LFS at passage 6 were used for these experiments. The expression levels were normalised against housekeeping genes GAPDH and ACTB and relative quantitation carried out as per Michael Pfaffl’s formula (Pfaffl 2001) (Chapter 3, Section 3.2.6), using the normal fibroblast cell line MRC5 as the control. The relative quantitation was also confirmed by using the REST© 2005 program. The log2 transformed ratios obtained in REST©2005 for the two cell lines were statistically compared using a t-test (Table 6.1). For selected candidate genes, the expression was also compared to the well characterised fibrosarcoma cell line HT1080. A number of the candidate genes were selected on the basis of a previous report of signature characteristic of GISTs (Allander, Nupponen et al. 2001). This included the genes TNFRSF6B, PRKCQ and GPR20, introduced in Chapter 5, Section 5.2.3.6. The former was expressed more highly in the GIST-M cell line, compared to LMS-LFS, MRC5 and HT1080 (Figure 6.6(a) and (b)). This was evident both on the comparative Ct method, as well as visually, on 3% agarose gel electrophoresis.

~~259~~

~~Chapter 6: GIST-M Cell Line~~

Figure 6.5 KIT mutation analysis (a) A portion of the DNA sequence analysis of exon 11 of the KIT sequence in the GIST-M cell line is shown in forward and reverse direction and (b) shows a negative control sample for comparison. The blackened square of the nucleotide adenosine, A, identifies the point of insertion of the mutation. Background wild type sequence is also present in the sample. An in frame insertion was detected in both the cell line and the source tumour in exon 11, described as 1720_1725 ins GCCCCA. This corresponds to an insertion of the amino acids serine and proline, or 573_574 ins SP. The insertion occurs within a codon as indicated by the position of the red arrow below (c), which shows a portion of the KIT exon 11 sequence. The numbers above the nucleotides correspond to the nucleotide number followed by the codon number. The amino acids encoded by each codon are shown beneath the corresponding codon. No mutations were detected in exons 9, 13 and 17 or in PDGFRA exon 18. W\XW [ZWW\[W [[W ! &9,=&99=(9,=&((=9&(=&&&=9&9=99&=(&,=&&&=(((=&9,=9&9=,&&=,9&=(&,=9,,=&&,=,99=,99==

~~2*9=.1*=1*:=9-7=9>7=1>8=9>7=1*:=,13=1>8=574=2*9=9>7=,1:=;&1=,13=975=1>8=;&1=;&1==~~

~~+02+2/0+======+1++2/1+======~~

~~,&,=,&,=&9&=&&9=,,&=&&(=&&9=9&9=,99=9&(=&9&=,&(=((&=&(&=(&&=(99=((9=9&9=,&9=(&(==~~

~~,1:=,1:=.1*=&83=,1>=&83=&83=9>7=;&1=9>7=.1*=&85=574=9-7=,13=1*:=574=9>7=&85=-.8==~~

~~+1.+2/2+======+11+2/3+======~~

~~&&&=9,,=,&,=999=(((=&,&=&&(=&,,=(9,=&,9=999=,,,=&&&=&((=(9,=,,9=,(9=,,&=,(9=99(=~~

~~1>8=975=,1:=5-*=574=&7,=&83=&7,=1*:=8*7=5-*=,1>=1>8=9-7=1*:=,1>=&1&=,1>=&1&=5-*=~~

~~260~~

~~Chapter 6: GIST-M Cell Line~~

Contrary to the findings of Allander and colleagues, the level of expression of PRKCQ 35 was in fact highest in the leiomyosarcoma cell line LMS-LFS (Figure 6.6 and Table 6.1). GPR20 was expressed in low levels in all cell lines assessed.

Phosphodiesterase 2A (PDE2A) expression was also assessed, revealing highest levels of expression in the GIST-M cell line, based both on the relative quantitation carried out using the comparative Ct method as well as the REST 2005© analysis (Figure 6.6(a)). This was also evident from the higher amplitude of the melt peaks produced by the PDE2A PCR product for the GIST-M cell line (Figure 6.7), as well as on gel electrophoresis (Figure 6.6(b)). In the cell line GIST882, PDE2A and KIT were shown by Frolov and colleagues to decrease in expression with imatinib treatment (Frolov, Chahwan et al. 2003).

The expression of HMGB1 was found to be the highest in the GIST-M cell line, compared to the other sarcoma lines LMS-LFS (Figure 5.10(a)) and HT1080, although the difference was not significant. The melt curve analysis for HMGB1 PCR products correspondingly showed a higher amplitude peak for the GIST-M cell line than for the LMS-LFS cell line (Figure 6.7(b)).

The expression of a number of receptor tyrosine kinases were also investigated in the characterisation of the GIST-M cell line. Platelet derived growth factor receptors A and B (PDGFRA and PDGFRB) were examined, as well as vascular endothelial growth factor receptor type 2 (VEGFR2 or KDR). PDGFRA and KDR are located on chromosome 4q11-q12, the same chromosome region as KIT.

~~35 PRKCQ expression will be discussed in greater detail in the characterisation of the LMS-LFS cell line (Chapter 7) 261~~

~~Chapter 6: GIST-M Cell Line~~

~~Relative Gene Expression in GIST-M and LMS-LFS Cell Lines~~

~~10 Relative ExpressionRelative Ratio~~

~~0 GIST-M LMS-LFS Genes of Interest~~

~~HMGB1 PDGFRA PDGFRB PDE2A PRKCQ TNFRSF6B VEGFR2 ACVR2B MAP4K4 GPR20~~

Figure 6.6 Expression of Biomarkers for GIST-M and LMS-LFS cell lines Transcript expression was examined by real time RT-PCR. (a) Expression is shown relative to the expression in the normal fibroblast cell line MRC5. Primer efficiency is taken into account and the values are normalised against the housekeeping genes GAPDH and βββ-actin using the REST 2005© program. Of note, GIST-M showed a high level of expression of phosphodiesterase 2A (PDE2A) while LMS-LFS showed a high level of expression of protein kinase C theta (PRKCQ). G-protein coupled receptor 20 (GPR20) levels of expression were generally low in both cell lines. All other potential biomarkers investigated, including tumour necrosis factor superfamily 6B (TNFRSF6B) and the receptor tyrosine kinases PDGFRA, PDGFRB and VEGFR2/KDR had higher levels of expression in the GIST-M cell line. (b) Selected PCR products were also run on a 3% agarose gel, comparing GIST-M, LMS-LFS and the well characterised fibrosarcoma cell line HT1080. For each gene, GIST-M is in the left hand lane (Lanes 1, 4, 8 and 11), LMS-LFS appears in Lanes 2, 5, 9 and 12 and HT1080 in Lanes 3, 6, 10 and 13. The 1kb 100 bp DNA ladder is in Lane 7.

~~PDE2A PRKCQ TNFRSF6B VEGFR2~~

~~1 2 3 4 5 6 7 8 9 10 11 12 13~~

~~262~~

~~Chapter 6: GIST-M Cell Line~~

~~PDE2A expression~~

~~GIST-M~~

~~LMS-LFS~~

~~NTC~~

~~GIST-M HMGB1 expression LMS-LFS β-actin~~

~~HEK293~~

Figure 6.7 Expression of Phosphodiesterase 2A (PDE2A) and High mobility group box 1 (HMGB1) The melt curves for these PCR products on real time RT-PCR (ABI PRISM ® 7700 Sequence Detection System, PE Applied Biosystems) are shown for the two cell lines GIST-M and LMS-LFS. NTC refers to the non-template control run for each target gene in every experiment. In graph (b), the expression in the immortalised human embryonic kidney cell line HEK293 is shown for comparison, exhibiting lower levels of expression for HMGB1 than the two newly characterised malignant cell lines. The melt curve for the housekeeping gene beta-actin is also shown. Both PDE2A and HMGB1 were expressed at higher levels in the gastrointestinal stromal tumour cell line GIST-M than the leiomyosarcoma cell line LMS-LFS.

~~263~~

~~Chapter 6: GIST-M Cell Line~~

Imatinib, which has been shown to achieve partial response rates as high as 80-90% in the treatment of GISTs (Section 5.3.4.1.4), is also known to have activity against PDGFRA and PDGFRB. Overall, the levels of expression of these target genes in both the GIST-M and LMS-LFS cell lines were low. However, the expression was significantly higher in the GIST-M cell line for all three of these receptor tyrosine kinases (Figures 6.6 and 6.8 and Table 6.1).

~~Table 6.1 Transcript Expression in GIST-M and LMS-LFS Cell Lines~~

# ,/-%T-6.0#11'-,( 2'- &#,# &3V /V/" 8 1-*32#24 *3# 8"(312#".4 *3# #$@ VSTSZW WTSRU VWT[WY RTRSV X UTZTW RTZWT VWTR[Y RTRSV /# RTVZV VYTVRS TRTYRT RTRUS 2/#/\ RTXVZ VTTWWX SZTTXY RTRUW 0/#X VSTYT[ VVTYUR SRTRSW RTRXU #XV VUTRSU VVTTRS XTZZY RTR[T /# VRTWVS VSTYRU XTTUT RTSRS W RTV[[ VRTW[[ UT[V[ RTSWZ

~~264~~

~~Chapter 6: GIST-M Cell Line~~

~~PDGFRA expression~~

~~GIST-M β-actin LMS-LFS~~

~~NTC~~

~~KDR (VEGFR2) expression GIST-M MRC5 LMS-LFS~~

~~NTC~~

Figure 6.8 Expression of PDGFRA and VEGFR2 The melt curves for (a) Platelet derived growth factor receptor A (PDGFRA) and (b) Vascular endothelial growth factor receptor type 2 (VEGFR2, also known as KDR, kinase insert domain receptor) are shown for the two cell lines GIST-M and LMS-LFS. NTC refers to the non-template control. In the lower graph, the melt curve of KDR (VEGFR2) for the normal fibroblast cell line MRC5 is shown. A small proportion of GISTs are known to have mutations in PDGFRA. Both the above genes encode, like KIT, receptor tyrosine kinases. The gastrointestinal stromal tumour cell line GIST-M exhibited the highest levels of expression of these genes.

~~265~~

~~Chapter 7: LMS-LFS Cell Line~~

~~6.2.8 Analysis of Telomere Maintenance Mechanism~~

The limited proliferative lifespan of cells, or the “Hayflick limit” is thought to consist of 50-60 population doublings, after which senescence ensues (Dhaene, Van Marck et al. 2000; Reddel 2000). This is due, in part, to the progressive shortening of telomeres. Telomeres shorten with each cell division, by 40-200 bp, to a critical point at 5-8 kb. Some malignant cells possess a telomere maintenance mechanism, allowing them to escape senescence. In most epithelial tumours, this involves the enzyme telomerase (Shay and Bacchetti 1997), a complex which enables the addition of TTAGGG tandem repeats at the telomeres. A greater proportion of mesenchymal tumours and some immortalised cell lines utilise an alternate mechanism for lengthening of telomeres, or ALT (Bryan, Englezou et al. 1997). The precise mechanism(s) involved in ALT is/are not known, but is thought to involve recombination mediated DNA replication (Henson, Neumann et al. 2002). The GIST-M cell line was investigated for the presence of ALT and telomerase.

~~6.2.8.1 TRAP assay for telomerase activity~~

Cell lysate was used with the primers, polymerase, dNTPs and TRAP buffer for the PCR reaction at the specified cycle parameters, as described in Appendix V. The PCR product was then run on a run on a 10 % non-denaturing polyacrylamide gel and stained with SYBR Green (Molecular Probes) for visualisation. If telomerase activity was present, the addition of TTAGGG tandem repeats would have occurred and a 6 bp ladder of products is visualised on the gel. Lysis buffer was used as the negative control and an osteosarcoma cell line MG-63 was used as the known positive control. The results of the TRAP assay for both the GIST-M cell line at passage 11 and the LMS- LFS cell line at passage 30 are shown in Figure 6.9. Neither of the above cell lines was shown to possess telomerase activity.

~~266~~

~~Chapter 7: LMS-LFS Cell Line~~

*71'1&3$$#0 V##***',# * V*# #***',# #VXU 2# 2 , !2'4 2'-,

0'+#0"'+#01 S TUVW X YZ [SR SS ST SU SV Figure 6.9 The Telomerase repeat amplification protocol (TRAP) assay Each row of numbers along the top of the gel refers to the number of μμμl of sample used at 1 μμμg/μμμl. The top row is the lysis buffer (negative control). The 2nd row is the PCR product from the GIST- M cell line (Lanes 2-5 and 1 μμμl in Lane 13 to spike the positive control). The 3rd row is the LMS- LFS cell line (Lanes 6-9 and 1 μμμl in Lane 14). The 4th row details the amount of the positive control osteosarcoma cell line MG-63 used in Lanes 10-14. The bottom row indicated where heat has been used to inactivate telomerase (Lanes 5, 9 and 12). If telomerase is present, as in the cell line MG-63, the 6 bp ladder of amplified telomeric TTAGGG repeats is electrophoresed on the gel, as in Lanes 10, 11, 13 and 14. The products disappear in Lane 12 where heat has been used to inactivate telomerase. These results show that unlike MG-63, both the GIST-M and LMS-LFS cell lines to not express telomerase.

~~6.2.8.2 ALT associated PML bodies (APB) immunostaining~~

Promyelocyte leukaemia (PML) nuclear bodies are present in the nuclei of many different cell types. They are thought to be involved in many vital cellular functions such as cell cycle regulation, differentiation, inflammatory responses and tumour suppression (Zhong, Salomoni et al. 2000). APBs are PML bodies that contain telomeric DNA and other telomere-specific DNA-binding proteins such as TRF1 and

~~267~~

~~Chapter 7: LMS-LFS Cell Line~~

~~TRF2. The presence of APBs is thought to be a hallmark of ALT (Yeager, Neumann et al. 1999).~~

Immunostaining for APBs was performed on cells cultured on chamber slides. The characteristic feature of APBs is the “donut” appearance of the colocalised PML nuclear body and superimposed telomeric DNA within the nuclei. APBs are present in only 5% of the cells at any given time, implying that ALT expression is manifested only at certain phases of the cell cycle (Henson, Hannay et al. 2005). If APBs are present in ≥10 cells or ≥0.5% of cells, it is considered a positive result. The GIST-M cell line was not shown to have APBs (Figure 6.10). The PML bodies and telomeric DNA did not colocalise in concentric rings as would be expected for an ALT+ cell line.

Figure 6.10 Alternate lengthening of telomeres (ALT) mechanism The Gastrointestinal stromal tumour cell line GIST-M was examined for ALT-associated PML bodies (APBs). APBs are thought to be involved in the ALT mechanism itself. These two images show individual cell nuclei, where, DAPI is used to stain the nuclei blue. Within the nuclei, the Promyelocytic leukaemia (PML) bodies are stained red and Telomeric DNA stained green. For a positive result, colocalisation of PML bodies and telomeric DNA within the nucleus in almost a concentric or “donut” fashion is required. The telomeric DNA in the APB should also have more intense fluorescence than the telomeres on the slide. The camera exposure time used in the evaluation of the slides was adjusted such that telomeres would not be visible relative to the telomeric DNA. Unlike the LMS-LFS cell line (shown in Figure 5.25), the GIST-M cell line does not exhibit this pattern. It therefore cannot be said to have APBs. This implies either that the cell line does not utilise ALT (alternate mechanism for lengthening of telomeres) as a telomere maintenance mechanism, or that it is yet to activate this mechanism.

~~268~~

~~Chapter 7: LMS-LFS Cell Line~~

~~6.3 DISCUSSION~~

In this study, a new GIST cell line was established and characterised. Although two other GIST cell lines (GIST882 and the short term culture GIST544) have been described in the literature, the full characterisation of these has not been published. The only other GIST cell line, GIST-T1, also derived from a metastatic deposit, has been examined more fully, with conventional karyotyping, comparative genomic hybridisation and a preliminary mutational analysis being carried out (Taguchi, Sonobe et al. 2002).

GIST-M is a unique cell line in that it was derived from a primary GIST, and one that was resected following six months of neoadjuvant treatment with imatinib. Even though the patient was observed to have a tumour that responded to Imatinib treatment, with a documented response on 18-FDG PET scanning, the tumour did not decrease in size and the patient’s symptomatology had not changed. Despite the documented response on PET scanning, at resection, there was a significant amount of viable tumour present. This raised the possibility of the cell line carrying both a primary mutation as well as an acquired secondary mutation.

The patient subsequently did develop recurrent disease, which was again treated with Imatinib. A follow up PET scan revealed no uptake. This would admittedly indicate that the tumour remained responsive to the drug. However, it is possible that there existed within the tumour cell population, a clone possessing a second mutation, given the recurrence of the disease after initial treatment.

The morphology, immunocytochemistry profile and KIT mutation of the cell line suggests that it retains the characteristics of the original tumour. Further characterisation of the cell line was carried out by conventional karyotyping, examination of biomarkers by real time RT-PCR and assessment of telomere maintenance mechanisms.

~~6.3.1 Immunophenotype~~

The majority of the markers examined are not specific for GISTs alone. Apart from the 80-90% positivity for KIT staining, studies to date estimate that 60-70% of GISTs express CD34, although this is more commonly seen in colorectal or oesophageal

~~269~~

~~Chapter 7: LMS-LFS Cell Line~~

GISTs, while 30-40% express smooth muscle actin (SMA) (Coindre, Trojani et al. 1986; Fletcher, Berman et al. 2002). The latter is more common in GISTs arising from small bowel. A small proportion (1-5%) display immunoreactivity for S100 and desmin. Given that the source tumour in this case was one that appeared to arise from the serosa of the duodenum, it is not unexpected to find that both the tumour and the cell line did not express CD34. Both tumour and cell line stained positively for desmin and vimentin and negatively for S100 and SMA, again providing evidence for the cell line maintaining the immunophenotype of the source tumour.

~~6.3.2 KIT expression~~

Early passages of the GIST-M cell line displayed strong expression for the KIT protein, a hallmark of GISTs. Later passages of the cell line subjected to KIT staining using identical protocols, revealed reduced expression and eventually no expression of KIT.

~~Fletcher et al noted that in general, KIT immunopositivity was required to diagnose GISTs with “rare” exceptions such as (Fletcher, Berman et al. 2002):~~

~~1*8.438= <.9-= 9>5.(&1= (>94&7(-.9*(9:7&1= +*&9:7*8= 4+= 345= ':9= <-.(-= c__-&;*=~~

~~.3=7&7*=(&8*8=(*&8*)=94=*=57*88=635=):*=94=842*=+472=4+=(143&1=*;41:9.43`=~~

~~5*7-&58= +4114<.3,= 4538/1+= 9-*7&5>`___= 5:2478= .3= 9-*8*= *=(*59.43&1=~~

~~(&9*,47.*8=8-4:1)='*=1&'*1*)=85.3)1*=(*11=47=*5.9-*1.4.)=89742&1=3*451&82=~~

~~2489= (438.89*39= <.9-= 345_= 9-*9-*7= 47= 349= 635= 548.9.;.9>= 8-4:1)= '*=~~

~~7*6:.7*)=+47=&=).&,348.8=4+= 345=.8=247*=(439*39.4:8_=~~

In this GIST-M cell line, which was derived from a tumour that was resected after the patient had received six months of imatinib (STI 571/Gleevec®/Glivec®, Novartis, Basel, Switzerland) treatment, a non KIT expressing clone may have self-selected and come to predominate. Similar findings have been observed by Heinrich and colleagues who comment that KIT expression appeared to be silenced in certain in vitro and murine GIST models while the cells continue to proliferate (Heinrich, Rubin et al. 2002). Heterogeneity within the tumour cell population may explain the phenomenon in this cell line. We hope to ideally perform a repeat karyotyping of the cell line at the later

~~270~~

~~Chapter 7: LMS-LFS Cell Line~~

~~passage in the future, to enable comparison with the earlier passage and assess whether the two populations are clonally related.~~

In a study involving patients with disease progression despite imatinib treatment, it was observed that tumours from two of the 26 patients had totally lost KIT expression. In one but not both of these, the loss of KIT expression on immunohistochemistry was associated with a biallelic loss of KIT loci (Debiec-Rychter, Cools et al. 2005).This raises the possibility of other unidentified activating mutations, either involving KIT signalling pathways, or perhaps other entirely unrelated mechanisms leading to escape from receptor dependence.

~~6.3.3 KIT mutation analysis~~

An early study found 89% of the tumours expressed KIT protein but that only 57% of the GISTs were found to have KIT mutations (Taniguchi, Nishida et al. 1999). Another study found similar rates of KIT mutations (Lasota, Jasinski et al. 1999). These studies, however, only examined exons 11 and 17 of the KIT gene. Up to 95% of GISTs have since been reported to harbour KIT mutations (Duensing, Heinrich et al. 2004). The majority of these involve exon 11 mutations (Figure 6.11(b)). Deletions and insertions are more prevalent in codons 557 to 559, whereas point mutations are known to affect codons 557, 559, 560 and 576 (Corless, Fletcher et al. 2004). The juxtamembrane (JM) region encoded by exon 11 functions as an autonomously folding domain that binds directly to the amino-terminal (ATP binding portion) lobe of the KIT kinase and inhibits KIT receptor activation. Mutation in the JM region prevents this autoinhibition, resulting in faster activation times (Chan, Ilangumaran et al. 2003).

Less commonly, GISTs are known to have mutations involving the KIT extracellular domain (exon 9) or the split tyrosine kinase domains (exons 13 and 17). Exon 9 mutations typically take the form of in frame duplications (Lux, Rubin et al. 2000; Heinrich, Corless et al. 2003a) but the mechanism by which these mutations result in KIT receptor activation is unclear. An exon 13 mutation resulting in a single amino acid substitution (Glu at Lys642) and KIT activation has also been described (Lux, Rubin et al. 2000). Studies suggest KIT mutations are acquired before other cytogenetic

~~271~~

~~Chapter 7: LMS-LFS Cell Line~~

~~abberations such as loss of 14 & 22 or 1p deletion (discussed below) during tumorigenesis.~~

A proportion (35% ) of the GISTs that do not express constitutively activated KIT are now known to have PDGFRA mutations, involving either exon 12 or more commonly, the activating loop (exon 18) (Heinrich, Corless et al. 2003b). As with KIT, these oncogenic mutations also result in PDGFRA activation.

Most studies of KIT protein expression and KIT mutation status in GISTs have also noted a small number of cases with either KIT protein expression but no detectable KIT mutation, or no KIT expression together with wild type KIT. This implies that other mutations may also be oncogenic in these tumours, for instance mutations affecting downstream signal transducers on the KIT pathways, mutations on the non-coding regions of KIT that may control its transcription or splicing or mutations affecting KIT promoter regions. Mutations may also cause inactivation of KIT-inhibitory phosphatases, up-regulation of the KIT ligand or KIT heterodimerisation with other activated receptor-tyrosine kinase proteins. With the emergence of resistance to Imatinib the clinical implications of these mutations have been examined further. A recent randomised trial of Sunitinib malate36 (SU11248, Pfizer, La Jolla,CA), in imatinib resistant patients with GISTs encouragingly revealed a significant delay in disease progression in the treatment group compared to those receiving placebo (Demetri, van Oosterom et al. 2006). Acquired secondary KIT and PDGFRA mutations have been described (Debiec-Rychter, Cools et al. 2005; Loughrey, Waring et al. 2006; Tornillo and Terracciano 2006). So-called “polyclonal resistance” is attributed to sequential development of point mutations in exon 17 and 13.

~~36 Sunitinib malate, formerly known as SU11248, is a tyrosine kinase inhibitor targeting KIT, PDGFRA, PDGFRB, VEGFR1, VEGFR2, VEGFR3 and FLT3. 272~~

~~Chapter 7: LMS-LFS Cell Line~~

~~SCF SCF~~

~~Membrane~~

~~P P Substrate~~

~~P P~~

~~P P Split Tyrosine Kinase domain, phosphorylated~~

~~Ligand binding domain (SCF)~~

~~In frame duplication exon 9 (13%) Membrane In frame mutation exon11 (71%) JM~~

~~Point mutation exon13 (4%) TK1~~

~~Point mutation exon17 (4%) TK2~~

~~Cytoplasm~~

Figure 6.11 Structure of the KIT receptor and location of mutations (a) The transmembrane KIT receptor has extracellular domains for binding of the ligand, stem cell factor (SCF). Binding leads to autodimerisation of the receptor, which activates the complex by phosphorylation (P) of the intracellular split tyrosine kinase domains (TK1 and TK2). This allows binding and phosphorylation of substrates, triggering various downstream cell signalling cascades. (b) Mutations in the KIT gene typically result in SCF independent constitutive activation of the KIT receptor. In a study of 48 GISTs, Rubin et al documented the frequency and distribution of KIT mutations, as given by the percentages above. 92% of GISTs in their study had KIT mutations (Rubin, Singer et al. 2001), but the frequency reported in the literature ranges from 21-92%. The GIST-M cell line was found to have an in frame insertion at exon 11, which encodes the juxtamembrane (JM) domain. The normal functions of the JM domain are thought to include the formation of an alpha helix, which inhibits the kinase domain and the provision of binding sites for proteins which inhibit KIT phosphorylation. These figures were adapted from Duensing et al and Heinrich et al (Heinrich, Rubin et al. 2002; Duensing, Heinrich et al. 2004).

~~273~~

~~Chapter 7: LMS-LFS Cell Line~~

The KIT mutation found in the cell line GIST-M is an in frame insertion of 6 nucleotides at exon 11, codon 573. While insertion mutations, which are rare, have been described (Wardelmann, Neidt et al. 2002), the particular mutation present in this cell line and its source tumour has not been reported before. Based on the KIT protein expression in the original tumour and early passages of the cultured cells, this mutation can be said to be an activating mutation. No additional mutations were found in exons 9, 13, 17 or exon 18 of PDGFRA. At the time this analysis was performed, these were the exons routinely examined for mutations in GISTs. It is possible that other secondary mutations occur at exons other than the above or downstream on the KIT and PDGFRA signalling pathways. Given that the cell line, in later passages lost KIT expression, it is possible that other KIT-independent mutation(s), as described above, may have occurred. Certainly, the multiple clonal aberrations noted in the karyotype of this cell line would suggest that the tumour and/or the derived cell line had acquired new mutations in its progression.

~~6.3.4 Cytogenetics~~

Cytogenetic aberrations noted to be associated with GISTs include monosomy of chromosomes 14 and 22, loss of 9 or 9p, complete or partial loss of 1 or 1p, loss of 15, loss of 3 or 3p, loss of 13q, loss of 10q, complete or partial gain of 5, and gain of chromosome 4 (Gunawan, Bergmann et al. 2002; Heinrich, Rubin et al. 2002). The most common aberrations in addition to monosomy of 14 and 22 are loss of 9 or 9p, 1 or 1p and 15. Chromosome 9 aberrations most often involve the 9p21 region, which contains tumor suppressor genes such as p16. It is surmised, that this may play a role in tumor progression and metastatic capability.

Three or more aberrations have been associated with high grade or recurrent GISTs, with 2.6 chromosomal aberrations noted in benign GISTs, 7.5 in malignant primary GISTs and 9 in metastatic GISTs (El-Rifai, Sarlomo-Rikala et al. 2000). Similarly, another group of low-risk GISTs were found to have a mean of 2.7 DNA copy number changes (range 0–5), including 0.6 gains (range 0–2) and 2.1 losses (range 0–5) (Langer, Gunawan et al. 2003). Progressive tumours had more than five changes and exhibited gains at 5p and 8p, and losses in 10q, 15q and 9p.

~~274~~

~~Chapter 7: LMS-LFS Cell Line~~

This cell line GIST-M exhibited multiple clonal anomalies. In line with the published literature, there was monosomy of chromosomes 14 and 22. In addition, there were deletions involving chromosome regions 1p, 9p, 10q and 15, which have also been previously reported in GISTs. These secondary aberrations alone, would be sufficient to categorise this cell line as a high grade malignant GIST. The cells also exhibit additional translocations, indicative of mutations accumulated during malignant transformation and disease progression. The tumour suppressor gene CDKN2A (p16INK4A) is located on chromosome 9p and has shown to be inactivated in malignant GISTs (Schneider-Stock, Boltze et al. 2003). Target genes at the other frequently involved chromosomal loci are yet to be elucidated.

~~6.3.5 Telomerase and ALT~~

Although the concept of immortalisation and malignancy has been investigated in epithelial tumours and to a lesser degree, in mesenchymal tumours, there has been no extensive evaluation of telomerase and ALT in GISTs.

In a small study evaluating the prevalence of telomerase in GISTs, 7 of 24 primary tumors and 5 of 5 metastatic tumors were found to have telomerase activity (Sakurai, Fukayama et al. 1998). Tumours expressing telomerase showed a significantly higher rate of proliferation. A higher rate of telomerase activity has been found in GISTs (67%) than in leiomyosarcomas (LMS) (18%) and malignant peripheral nerve sheath tumours (MPNST) (48%) (Gunther, Schneider-Stock et al. 2000). Interestingly, there was no detectable telomerase activity in 4 of the 21 GISTs studied. The alternate lengthening of telomeres (ALT) mechanism was not, however, investigated in this or the previous study.

The other salient finding was that telomerase activity was detected in a recurrent GIST while the primary tumour in the same patient had not been found to express it. It was proposed that telomerase activity occurred during the progression of malignant GISTs. This is consistent with a review of telomerase activity in all types of tumours, in which telomerase activity was detected only in late-stage tumours, even though levels of the telomerase RNA component hTERC were upregulated in the early pre-neoplastic stages

~~275~~

~~Chapter 7: LMS-LFS Cell Line~~

~~(Dhaene, Van Marck et al. 2000). In other words, telomerase activation occurs late in the evolution of the malignancy.~~

The GIST-M cell line was investigated for both telomerase and ALT activity and found to express neither, based on the results of the TRAP assay and APB immunostaining. In other words, a telomere maintenance mechanism (TMM) mechanism was not detected. This may indicate that either that telomerase is yet to be activated in this cell line, or that is the cells are still mortal. Even if GIST-M is mortal, and goes on to senesce with repeated passaging, it has potential value as a short term culture, as demonstrated by previous studies utilising short term GIST cultures (Duensing, Medeiros et al. 2004; Prenen, Cools et al. 2006).

~~6.3.6 Expression of Other Receptor Tyrosine Kinases~~

Approximately 35% of GISTs with wild type KIT have been shown to harbour PDGFRA mutations (Heinrich, Corless et al. 2003b). These GISTs express high levels of PDGFRA protein, however, some GISTs with KIT exon 11 mutations are also known to express PDGFRA but at lower levels (West, Corless et al. 2004). This is borne out in the GIST-M cell line, which expressed higher levels of PDGFRA on qRT-PCR than the other sarcoma cell lines it was compared to, namely LMS-LFS and HT1080. Interestingly, another recent gene expression study found PDGFRA expression in GISTs to be dependant on anatomic site, with over-expression seen in gastric GISTs rather than intestinal GISTs (Antonescu, Viale et al. 2004).

The expression of another receptor tyrosine kinase, VEGFR2 or KDR (kinase insert domain-containing receptor) was also examined in the GIST-M cell line. VEGF is known to be involved in angiogenesis, specifically endothelial cell proliferation and migration. Hence one of the goals of anti-angiogenic therapies has been to target VEGF or its receptor, with many of these in current pre-clinical and clinical trial stages (Angelov, Salhia et al. 1999; Laird, Vajkoczy et al. 2000; Mendel, Laird et al. 2003; Desai, Maki et al. 2004; Ryan and Wedge 2005). The expression of KDR in the GIST- M cell line was greater than in the LMS-LFS and HT1080 cell lines, identifying KDR

~~276~~

~~Chapter 7: LMS-LFS Cell Line~~

~~both as a potential biomarker, as well as a therapeutic target. However its expression was low compared to that of PDE2A (see below).~~

~~6.3.7 Expression of KIT-associated Biomarkers~~

A number of previous gene profiling studies had identified genes upregulated in GISTs, compared to other soft tissue sarcomas. These included markers that were examined in the present study by real time RT-PCR, such as G-protein coupled receptor 20 (GPR20), tumour necrosis factor receptor superfamily 6B (TNFRSF6B), phosphodiesterase 2A (PDE2A) and protein kinase C theta (PRKCQ) (Allander, Nupponen et al. 2001; Nielsen, West et al. 2002; Frolov, Chahwan et al. 2003; Antonescu, Viale et al. 2004; Subramanian, West et al. 2004). In the present study, expression of PDE2A and TNFRSF6B were significantly upregulated in GIST-M, compared to the leiomyosarcoma cell line whereas there was significantly less expression of PRKCQ.

TNFRSF6B, also known as Decoy receptor 3 (DcR3) is known to bind Fas ligand (FASL) and inhibit FASL induced apoptosis (Pitti, Marsters et al. 1998). This is utilised by natural killer cells and cytotoxic T lymphocytes against virus infected and malignant cells. TNFRSF6B has previously been reported to be overexpressed in lung cancers, as well as adenocarcinomas of the gastrointestinal tract, implicating a role in tumour progression (Pitti, Marsters et al. 1998; Bai, Connolly et al. 2000). It may play a similar role in GISTs. The cell line GIST-M showed higher levels of expression of TNFRSF6B than the other cell lines studied, in keeping with other gene expression studies of GISTs.

It has been proposed that G protein coupled receptors (GPCRs) can cause transactivation of receptor tyrosine kinases (RTKs) such as EGFR and PDGFR, with phosphatidylinositol 3 kinase (PI3K) and protein kinase C (PKC) acting as early intermediates in this pathway (Luttrell, Daaka et al. 1999). GPR20, on the basis of recent gene expression studies identifying it as highly expressed in GISTs, is thought to play a role in the KIT pathway. Contrary to these studies, the expression of GPR20 in the GIST-M cell line was found to be low.

~~PKC proteins regulate KIT signalling through inhibitory phosphorylation of KIT interkinase domain serine residues, and through other downstream signalling~~

~~277~~

~~Chapter 7: LMS-LFS Cell Line~~

intermediates (Subramanian, West et al. 2004). Subramanian et al found that the KIT exon 9 and 11 mutant GISTs expressed higher levels of PRKCQ as compared to PDGFRA mutants. The GIST-M cell line demonstrated an exon 11 mutation. However, its expression of PRKCQ was lower than that of the leiomyosarcoma cell line LMS-LFS characterised in the subsequent sections of this chapter. This may, in part be due to the loss of KIT expression in the later passages of the cell line.

~~6.4 SUMMARY and PERSPECTIVES~~

A new GIST cell line derived from a primary retroperitoneal tumour was established and maintained in culture for 18 months. • It retained the immunophenotypic profile of the original tumour, with the exception that KIT expression was lost in the later passages of the tumour. • This was discussed in the context of an identified KIT exon 11 insertion mutation (1720_1725 ins GCCCCA), the first reported mutation of this type. • Conventional karyotyping revealed multiple clonal anomalies, such as monosomy of 14 and 22 which have been previously reported in GISTs, as well as translocations not previously reported. • There was high expression of phosphodiesterase 2A (PDE2A), which is known to be downregulated by Imatinib treatment and MAP4K4 was also overexpressed. There was high expression of phosphodiesterase 2A (PDE2A), which is known to be downregulated by Imatinib treatment and MAP4K4 was also over-expressed. The GIST-M cell line therefore contributes to the GIST and sarcoma cell line repository and will serve as an in vitro model to further delineate mechanisms of KIT activation and inactivation, including signal transducers such as PDE2A and members of the MAPK cascade.

~~278~~

~~Chapter 7: LMS-LFS Cell Line~~

~~2" +(~~

~~7. CHARACTERISATION of a NOVEL LMS-LFS CELL LINE from a LEIOMYOSARCOMA in a patient with LI FRAUMENI SYNDROME~~

~~279~~

~~Chapter 7: LMS-LFS Cell Line~~

~~7.1 INTRODUCTION~~

In 1969 Li and Fraumeni first described an over-representation of sarcomas among the siblings and cousins of five families on reviewing the records of 648 children with Rhabdomyosarcoma (RMS) (Li and Fraumeni 1969a; Li and Fraumeni 1969b). Pearson et al in 1982 designated this autosomal dominant condition Li-Fraumeni Syndrome (LFS) and described it as classically involving a proband with sarcoma, aged under 45 with any cancer in a first degree relative also under the age of 45 and another first or second degree relative with cancer under the age of 45 or sarcoma at any age (Pearson, Craft et al. 1982). It was not until 1990 however, that mutations in the p53 gene were linked to LFS (Malkin, Li et al. 1990).

It is now known that the mutations can span the entire gene but tend to cluster in the DNA-binding domains (Birch, Hartley et al. 1994; MacGeoch, Turner et al. 1995; Barel, Avigad et al. 1998; Bougeard, Limacher et al. 2001; Varley, Attwooll et al. 2001). Mutations in PTEN and CDKN2A have been excluded from the pathogenic process in LFS (Burt, McGown et al. 1999; Portwine, Lees et al. 2000). The clinical and molecular manifestations of the syndrome are reviewed in greater detail by Varley et al (Varley, Evans et al. 1997). The spectrum of tumours seen in LFS patients with germline p53 mutations is similar to those with sporadic p53 mutations and hence cell lines derived from these patients serve to provide useful models for the ongoing study of tumourigenesis. Earlier studies have either immortalised skin fibroblasts obtained from patients with LFS by using Aflatoxin B1, irradiation (Tsutsui, Fujino et al. 1995; Tsutsui, Tanaka et al. 1997). Other researchers used fibroblasts obtained from tissue adjacent to a breast tumour by using SV40 or HPV (Maclean, Rogan et al. 1994; Bryan, Englezou et al. 1995). A minority of cell lines derived from LFS patients has spontaneously immortalised (Rogan, Bryan et al. 1995) but none of these were derived from sarcomas.

~~280~~

~~Chapter 7: LMS-LFS Cell Line~~

In the following sections, the characterisation of a unique cell line from tissue obtained from a sarcoma in a patient with LFS is described. The methods used are as described in Section 5.2 and in greater detail in Appendix V.

~~7.2 RESULTS~~

~~7.2.1 Tumour Source~~

~~The clinical details of the 37 year old patient, the histology of the tumour and the family pedigree (Figure 7.1) are presented in Appendix V.~~

~~20yrs ?Liver Ca~~

~~43yrs Fibrosarcoma~~

* * * KS, 37yrs 8yrs DCIS, Glioma, LMS Adrenal

~~MS, 3yrs RMS~~

Figure 7.1 Pedigree of Li Fraumeni syndrome (LFS) family KS: The leiomyosarcoma (LMS) from this patient was used to establish the cell line. The patient had previously had bilateral mastectomies for Ductal carcinoma in situ (DCIS) and a craniotomy for a Glioma. The 3 year old daughter had a lower limb Rhabdomyosarcoma (RMS) resected. Squares represent males and circles, females. Solid black colouring refers to affected family members and a line across the circle or square refers to deceased individuals. The * refers to those who declined to be tested for germline p53 mutation.

~~7.2.2 Establishment of Cell Line~~

~~When the cell pellet was initially seeded to flasks, growth was slow, with the tumour cells forming distinct colonies or nests in between fibroblasts. These were allowed to~~

~~281~~

~~Chapter 7: LMS-LFS Cell Line~~

grow to a greater degree of confluence, at which point quick trypsinisation was carried out to select for enriched tumour cell populations. Fibroblasts decreased in number sufficiently after repeating this procedure a second time. By passages 7-10 the growth rate of the cells also increased. The doubling time of the cells was 44 h.

~~7.2.3 Morphology~~

As examined by light microscopy (Figure 7.2), the cells appeared spindle shaped to polygonal, with short cytoplasmic processes growing loosely and in nests. The pleomorphic nuclei are large with prominent nucleoli and the cytoplasm is granular.

~~7.2.4 Immunocytochemistry and Immunohistochemistry~~

The results of the immunostaining is summarised in Table 7.1. Strong staining was seen for CD10 in both the cell line and the original tumour and there was patchy positive staining in both for CD99 (Figure 7.3). There was patchy weak staining for smooth muscle actin (SMA). The tumour showed strong staining for Cytokeratin AE1/AE3 (Figure 7.3), which is the only marker that did not correlate with the cultured cells. Markers that were negative for both the original tumour and primary culture included bcl2 & S100, while markers that were positive in both included desmin and the mesenchymal marker vimentin.

Immunostaining for p53 was negative for all but one of the 6 p53 antibodies used. The p53 antibodies map to different amino acid residues of the p53 protein, with Pab 240 being the only antibody mapping to the central DNA-binding region of the protein that is affected by the known mutation in this cell line (Pisters 2002; Lévy, Vidaud et al. 2004). Both the tumour and the cell line were negative for Pab 240. Antibody 1801, which maps to the initial amino portion of the protein did stain positive for the original tumour and the plasma clot of the cell line (Figure 7.4). There was positive staining for Kitenin (Figure 7.5) and negative staining for KAI1/CD82, in keeping with the inverse relationship previously observed in epithelial tumours (Lee, Park et al. 2004).

~~282~~

~~Chapter 7: LMS-LFS Cell Line~~

Figure 7.2 Original Tumour and LMS-LFS Cell Line Morphology (a) Haematoxylin and Eosin (H&E) stain of the original tumour showing sheets of poorly differentiated tumour cells. The tumour was classified as high grade based on the nuclear atypia, pleomorphism and degree of differentiation. It was highly invasive, involving multiple organs in the retroperitoneum. (b) Phase contrast microscopy showing nests of tumour cells in culture at early passage after quick trypsinisation to enrich tumour cell population and exclude fibroblasts. (c) Phase contrast microscopy of LMS-LFS cell line in culture at passage 23, with no contaminating fibroblasts. (d) H&E stain of the cell line cultured on chamber slides. This cell line has been in continuous culture more than 12 months and had been passaged 40 times at the time of writing.

~~283~~

~~Chapter 7: LMS-LFS Cell Line~~

~~Table 7.1 Immunostaining results comparing original tumour and LMS-LFS cell line on chamber slides and as paraffin embedded plasma clots~~

Antibody Original Tumour Plasma Clot Cell Line* Chamber Slide* #1+', &&& < & :'+#,2', && < &&& *72-)#0 2',8-S 8-U && < & 8− < & *)'2 & < & SRR − < & * SR &&& < &&& * UV− < & * [[ && < && !*T− < − .WUSZRS && && − .WU 6S −−− .WU. TVR −−− .WU. VTS −−− .WU. SXTR −−− .WU 6Y −−− 78S * ZT −−− 7'2#,', &&& &&& &&& #+'/3 ,2'2 2'4# 11#11+#,2 -$ 12 ',',% ',2#,1'27S &&& ,2#,1# 12 ',',% ," 1YW$ !#**1 12 ',',% .-1'2'4#R && -"#0 2# .-1'2'4# 12 ',',% 5'2& SW−YW$ !#**1 1&-5',% .-1'2'4# 12 ',',%R & -/3'4-! * -0 $-! * 5# ) .-1'2'4# 12 ',',%R − <#% 2'4# 12 ', 0SR$ !#**1 .-1'2'4#R< <-2 #2#0+',#"T#3&#!#***',#5 1 11#11#" 1!& + #01*'"#!3*230#1 ," $-01-+#+ 0)#01Q31',% -2&!& + #01*'"#1 ,".* 1+ !*-21Q 1"#1!0' #"',2#62

~~284~~

~~Chapter 7: LMS-LFS Cell Line~~

~~! "~~

Figure 7.3 Selected results of immunostaining for LMS-LFS cell line on chamber slides Strong positivity for CD10 is shown in (a) and weak positive staining for Cytokeratin AE1/AE3 (b) is shown at 40X magnification. Strong positive staining for vimentin (c) and weaker staining for CD99 (d) is shown at 10X magnification. The alkaline phosphatase method was used for these antibodies, with fuschin as the chromogen. Positive staining is thus red. Other markers examined included desmin, SMA, S100, cKit, CD34 and bcl2, some of which are shown in the following figure. For most of the markers, tissue from normal appendix was used as the positive control. The exceptions were CD10 and bcl-2, where tissue from lymph nodes served as positive control, CD99, where Ewing’s sarcoma was the positive control and cKit where both the appendix and a known gastrointestinal stromal tumour served as the positive control.

~~285~~

~~Chapter 7: LMS-LFS Cell Line~~

Figure 7.4 Selected results of immunostaining for LMS-LFS cell line p53 staining for 5 of the 6 antibodies used was negative. Images (a) and (b) show the positive (brown) staining obtained using the antibody 1801 in the original tumour and the plasma clot of the cell line respectively. Interestingly, the chamber slide culture (c) did not stain positive. The p53 mutation in this tumour and cell line occurs in the central DNA binding region resulting in a truncated protein. Antibody 1801 maps to the initial amino portion of the p53 protein and hence may still bind to the truncated protein. The p53 antibody Pab 240 which maps to the central DNA binding region of the protein did not stain positive for either the original tumour or the cultured cells (not pictured).

Figure 7.5 Immunostaining for Kai-1 and Kitenin in the LMS-LFS cell line The above panel shows in (a) the negative control for Kitenin, (b), the positive brown cytoplasmic staining for the plasma clot of the cell line embedded in paraffin and (c) the positive cytoplasmic staining for the cells cultured on chamber slides. Haematoxylin is used for nuclear counterstaining (blue). The appearance of brown nuclear staining in the plasma clot is likely to be an artifact, resulting from the generation of the plasma clot itself, or the antigen retrieval required for the paraffin embedded tissue. This was not seen in the chamber slide cultures, which were fixed with acetone and did not require antigen retrieval. Immunostaining for Kai-1 was negative for the original tumour as well as the cell line (not pictured).

~~286~~

~~Chapter 7: LMS-LFS Cell Line~~

~~! "~~

Figure 7.6 Immunocytochemistry in LMS-LFS cell line Staining for bcl2 (a), S100 in (b), ckit in (c) and smooth muscle actin, or SMA in (d) is shown. The alkaline phosphatase method was used for these antibodies, with fuschin as the chromogen. Positive staining is thus red. Other markers examined included desmin and CD34 (not pictured). For most of the markers, tissue from normal appendix was used as the positive control. The exceptions were CD10 and bcl-2, where tissue from lymph nodes served as positive control, CD99, where Ewing’s sarcoma was the positive control and cKit where both the appendix and a known gastrointestinal stromal tumour served as the positive control. It is not unusual for a leiomyosarcoma to exhibit patchy or weak positivity for epithelial markers such as cytokeratin (as shown in the previous figure), or neural markers such as S100 in (b) above. The original tumour from which this cell line was derived was negative for SMA and the cell line shows very weak focal staining for this marker.

~~287~~

~~Chapter 7: LMS-LFS Cell Line~~

~~7.2.5 Cytogenetics~~

~~The cell line at passage 15 was karyotyped. Twenty metaphases were analysed and a composite karyotype constructed, as shown in Figure 7.7. Based on ISCN convention, the karyotype can be written as~~

/380-0-31`$$`8$`&))+6-,`8+`8+`&)),5,-`8,`8-`&)).6`= 8.`&))/5+/`&))/6++=,`&))05,-`80`(1`.+*6+*`8+*`= 8+*`&))+-5++`)*7+-9+bb+-6,-bb5++`8+.`)*7+0&))+05+-= .38+0b6+-b`8+0`&))+16,/`)*1+26,+=,`&))+36+-=,`= &))+35+-`(,*`.,,6+*`(2&7)*7$ (5+1 2= /380-`.)*2`&))+.6-, (5- 2+**8+*0`.)*2=, (5,* = The clone was near triploid with several unbalanced translocations involving the terminal bands of chromosomes 2, 6, 16, 17 and 19. There were losses of several chromosome regions, namely 1, 2, 3, 4, 5q, 6, 10, 14, 16, 18q21-qter, relative to the triploidy as well as gains in chromosomes 7 and 22. The regions interpreted as losses may however be present in the derivative chromosomes. There was isochromosome formation of 10q and 22q. Dicentrics, rings and telomeric association was also observed (Figure 7.8 (a) and (b)). The significance of telomeric anomalies and the TMM was hence investigated further by TRAP assay and APB and TRF2 staining

~~7.2.6 TP53 Mutation Analysis~~

The sequencing reaction revealed the presence of the 637 C>T mutation in the TP53 gene which results in a premature termination codon (Figure 7.9).This mutation is identical to that identified in the patient whose tumour this cell line was derived from, as well as her daughter. The TP53 databases maintained by the International Agency for Research on Cancer (IARC) and Institut Curie (http://www-p53.iarc.fr/ and http://p53.free.fr/ ) (Soussi, Dehouche et al. 2000; Olivier, Eeles et al. 2002; Soussi, Kato et al. 2005) report 325 reported mutations at codon 213, 236 of which refer to this C>T mutation at nucleotide 637, resulting in a stop codon. These mutations have been reported in Li Fraumeni syndrome, as well as in multiple other sporadic tumours, including lung and adrenocortical tumours. The biological activity of this mutant has

~~288~~

~~Chapter 7: LMS-LFS Cell Line~~

~~not been tested (Beroud and Soussi 1998; Beroud, Collod-Beroud et al. 2000; Soussi, Dehouche et al. 2000; Beroud and Soussi 2003) (http://www.umd.be).~~

~~289~~

~~Chapter 7: LMS-LFS Cell Line~~

Figure 7.7 Composite karyotype of the LMS-LFS cell line The cell line at passage 15 was karyotyped. The above shows the near triploid characteristic and multiple anomalies. Translocations, relative gains and losses are described in greater detail within the text. The clone was near triploid with several unbalanced translocations involving the terminal bands of chromosomes 2, 6, 16, 17 and 19. There were losses of several chromosome regions, relative to the triploidy as well as gains in chromosomes 7 and 22. The regions interpreted as losses may however be present in the derivative chromosomes. There was isochromosome formation of 10q and 22q.

~~290~~

~~Chapter 7: LMS-LFS Cell Line~~

Figure 7.8 Cytogenetic Analysis of LMS-LFS cell line (a) Dicentrics and (b) telomeric associations were also observed in the metaphases analysed. These features have been reported to be associated with shortening of telomeres and may still occur in immortal cell lines that maintain telomere lengths, avoiding senescence. Telomeric dysfunction can lead to unstable ring and dicentric chromosomes and anaphase bridges. This can cause chromosomal fragmentation through persistent bridge-fusion-breakage (BFB) events. The significance of telomeric anomalies and the TMM utilized by this cell line was hence investigated further by TRAP assay and APB and TRF2 staining. This is considered in greater detail in the discussion.

~~291~~

~~Chapter 7: LMS-LFS Cell Line~~

~~12345 6 7 8 9 10 11~~

~~Transactivation Proline-rich DNA binding Oligomerisation Regulation (1-42; 43-62) (63-97) (102-292) (323-356) (363-393) N- -C P P P P P P P Ac P P Ac IARC p53 database~~

*-"-,TSU*1332 2'-,

~~.0#+ 230#2-.3&8*-"-, 303,! 2#", !2'4#.WU.0-2#',~~

Figure 7.9 Schematic representation of the TP53 gene The majority of mutations occur in the central DNA binding region (exons 5-8). The mutation present in this LMS-LFS cell line is a nonsense mutation in exon 6 codon 213, resulting in a truncated p53 protein. The schematic representation of the TP53 gene was adapted from the International Agency for Research on Cancer’s p53 mutation database (http://www-p53.iarc.fr/index.html).

~~292~~

~~Chapter 7: LMS-LFS Cell Line~~

~~7.2.7 Analysis of TMM: TRAP Assay and APB staining~~

The cell line at passage 30 was used for assessment of telomere maintenance mechanisms. The results of the TRAP assay as shown in Figure 6.9 confirmed that the LMS-LFS cell line did not express telomerase. APB staining, however, was positive, indicating that this immortal cell line utilises ALT as its TMM (Figure 7.10).

Figure 7.10 ALT expression in LMS-LFS cell line. The LMS-LFS cell line was examined for alternate mechanism for lengthening of telomeres (ALT)- associated PML bodies (APBs). APBs are thought to be involved in the ALT mechanism itself. These four images show individual cell nuclei, where, DAPI is used to stain the nuclei blue. Within the nuclei, the Promyelocytic leukaemia (PML) bodies are stained red and telomeric DNA stained green. The colocalisation of PML bodies and telomeric DNA within the nucleus in almost a concentric or “donut” fashion (arrows), is the characteristic appearance of APBs and is pathognmonic of ALT+ cells. The LMS-LFS cell line is ALT+ and therefore immortal.

~~293~~

~~Chapter 7: LMS-LFS Cell Line~~

The LMS-LFS cell line at passage 30 was considered to be ALT+ as 10% of its nuclei stained positive for APBs. At present a paraffin embedded tumour specimen is thought to be ALT+ if 1 10 cells or 10.5% of the cells stain positive for APBs. In the case of cell lines, no cell lines known to be ALT+ have <5% of their nuclei positive.

~~7.2.8 Expression of Biomarkers on real-time RT-PCR~~

The expression of some of the biomarkers was considered in the previous section on the characterisation of the GIST-M cell line (Section 6.2.7). The LMS-LFS cell line was shown to express receptor tyrosine kinases such as KDR (VEGFR2), PDGFRA and PDGFRB but at lower levels than the GIST-M cell line (Figures 6.6 and 6.7). The differential expression of PDGFRA was statistically significant, with a p value of 0.031 compared with that of the GIST-M cell line (Table 6.1).

The expression of phosphodiesterase 2A (PDE2A) and high mobility group box 1 (HMGB1) was shown to be higher in the GIST-M cell line than LMS-LFS (Figure 6.7 and Table 6.1). Similarly, the expression of tumour necrosis factor receptor superfamily 6B (TNFRSF6B) was also higher in the GIST-M cell line than LMS-LFS. The higher expression of PDE2A and TNFRSF6B in the GIST-M cell line was statistically significant (Table 6.1).

Of greatest statistical significance was the higher expression of PRKCQ in the LMS- LFS cell line (Figure 7.11 and Table 6.1). This candidate gene has been shown in previous studies to be overexpressed in GISTs. However, in this project, the leiomyosarcoma cell line LMS-LFS was found to have a much higher expression of PRKCQ than the GIST-M cell line (p = 0.014). This is considered further in the discussion (Section 7.3.2).

~~294~~

~~Chapter 7: LMS-LFS Cell Line~~

~~/V/"~~

~~&3!#***',#~~

~~<3*.0'+#0~~

~~%-7T[U!#***',#~~

~~/V/" %3SRZR TRR . STUV W X Y Z [~~

Figure 7.11 PRKCQ expression in LMS-LFS cell line (a) Melt curve analysis for the gene protein kinase C theta (PRKCQ), with the highest peak seen for the LMS-LFS cell line. Two other cell lines, GIST-M and the immortalized embryonic kidney cell line HEK293 have less expression of this gene, as evidenced by the low amplitude of the peaks and the primer dimers (PD), at the same temperature as the PD seen for the non template control (NTC). (b) 3% agarose gel showing the PCR products for PRKCQ (216 bp), KDR or VEGFR2 (211 bp), GAPDH (212 bp), PDE2A (239 bp) and HMGB1 (193 bp) in Lanes 2-6 respectively in the LMS-LFS cell line, compared to the expression of PRKCQ, GAPDH and HMGB1 in the fibrosarcoma cell line HT1080 (Lanes 7-9). The 1kb 100bp quantitative ladder (arrow indicating 200bp) is in Lane 1, indicating that all the amplicon sizes are within the expected range and that the amount of PRKCQ expressed by LMS-LFS (Lane 2) is greater than that for HT1080 (Lane 7). VEGFR2: Vascular endothelial growth factor receptor 2, PDE2A: Phosphodiesterase 2A, HMGB1: High mobility group box 1, GAPDH: Glyceraldehyde-3-phosphate dehydrogenase.

~~295~~

~~Chapter 7: LMS-LFS Cell Line~~

~~7.3 DISCUSSION~~

~~7.3.1 LMS-LFS cell line compared to Original Tumour~~

In this study a new leiomyosarcoma cell line derived from a retroperitoneal tumour in a patient with LFS was established and characterised. The immunocytochemistry profile of the cell line suggests that it retains the characteristics of the original tumour. In addition, an inverse relationship between KAI1 and kitenin expression was demonstrated, which has not, to date, been examined in STS or sarcoma cell lines. The cell line was shown to carry the same TP53 mutation as had been established from the family members, with loss of heterozygosity of the remaining allele.

Further characterisation was carried out by karyotyping the cell line and examining the expression of potential prognostic biomarkers. The demonstation of the use of ALT as the TMM mechanism is consistent with LMS-LFM being an immortal cell line. This is the first report of a cell line derived from a sarcoma in a patient with LFS and hence a valuable resource for studies of tumourigenesis and tumour progression in the context of germline TP53 mutations.

~~7.3.2 Expression of PRKCQ on real time RT-PCR~~

~~The expression of PRKCQ was found to be greatly overexpressed in the LMS-LFS cell line, relative to MRC5 and compared to the GIST-M cell line.~~

Protein kinase C (PKC) is a family of serine- and threonine-specific protein kinases that can be activated by calcium and the second messenger diacylglycerol (DAG). PRKCQ, or protein kinase C θ, however, is thought to be a calcium-independent but DAG- dependent (Monnerat, Henriksson et al. 2004). PRKCQ is important for T-cell activation and interleukin 2 (IL2) production (Altman and Villalba 2003). It is required for the activation of the transcription factors NF-kappaB and AP-1 (Sun, Arendt et al. 2000). It is also known to be expressed at neuromuscular junctions of skeletal muscle and parts of the nervous system (Wilda, Ghaffari-Tabrizi et al. 2001). Thus it thought to play a role in signal transduction between nerve and muscle. Little is known about the role of PRKCQ in cancer, other than its noted over-expression in GISTs regardless of

~~296~~

~~Chapter 7: LMS-LFS Cell Line~~

the type of mutation (KIT or PDGFRA) (Blay, Astudillo et al. 2004; Duensing, Joseph et al. 2004). However, other members of the protein kinase C family are implicated in angiogenesis and tumour progression and metastasis. They (PKC α, β and ε) mediate VEGF-induced angiogenesis and are involved in the regulation of integrins (PKCα) (Griner and Kazanietz 2007).

The high expression of PRKCQ in the LMS-LFS cell line would seem to contradict the findings of the previous authors who determined it to be a highly sensitive and specific marker for GISTs. However the expression of this protein has not been extensively studied in other sarcomas (Blay, Astudillo et al. 2004). This is of note as PKC inhibitors active against other members of the PKC family are in Phase II and III trials for glioblastoma and non Hodgkin’s lymphoma, among others (Teicher 2006).

~~7.3.3 TP53 mutation~~

The TP53 exon 6 codon 213 mutation in this cell line, which has been described previously in LFS (Hainaut ; Frebourg, Barbier et al. 1995; Olivier, Eeles et al. 2002) falls within the commonly involved DNA-binding region but is less frequent than other mutations found in LFS families. This is a null mutation, resulting in a truncated p53 protein.

~~7.3.4 Cytogenetic aberrations in LMS-LFS~~

The genetic instability evident from the cytogenetic anomalies seen in this cell line, may be due to the loss of p53 function, which is known to be associated with such instability, resulting in changes in chromosome ploidy and gene amplifications (Fukasawa, Wiener et al. 1997; Carroll, Okuda et al. 1999). There is no literature specifically examining cytogenetic anomalies in sarcomas from patients with LFS.

This LMS-LFS cell line, derived from a high grade tumour, is near triploid and shows multiple other anomalies, in keeping with its germline TP53 mutation. The cell line does carry some of the features of the incomplete karyotyping done on the original tumour, such as the abnormal chromosome 17. Compared to the original tumour, the cell line had increased in ploidy, perhaps as a result of accumulating mutations.

~~297~~

~~Chapter 7: LMS-LFS Cell Line~~

The dicentrics and telomeric association noted on karyotyping is likely to also result from the genetic instability, leading to gene amplification. These features are markers of telomere shortening, which, in this immortal cell line, may imply that telomere lengths are not uniformly maintained (Tsutsui, Kumakura et al. 2003). As discussed in Chapter 1 (Section 1.8.3.2), telomeric dysfunction can lead to unstable ring and dicentric chromosomes and anaphase bridges. This can cause chromosomal fragmentation through persistent bridge-fusion-breakage (BFB) events leading to a continuous reorganisation of the tumor genome (Mandahl, Heim et al. 1988; Gisselsson, Jonson et al. 2001; Tsutsui, Kumakura et al. 2003). Recent literature has begun to evaluate the cytogenetics characteristics of soft tissue sarcomas with complex karyotypes in relation to prognostics37. Mertens et al studied 400 benign and malignant soft tissue tumours and found breakpoints in regions 1p1, 1q4, 14q1 and 17q2 and gain of regions 6p1/p2 to be independent predictors of adverse outcome (Mertens, Stromberg et al. 2002). An increasing effect on metastatic risk was seen with increasing involvement of the selected cytogenetic variables, even when different histopathological types were studied and there was a positive correlation between tumour grade and number of aberrations (Mertens, Fletcher et al. 1998).

Analyses of leiomyosarcomas (LMS) in particular have revealed clonal rearrangements involving 1q21, 2q21, 3q21, 9q, 10q22, 11p15, 11q21, 14q24, and 20q13.1, often in unbalanced translocations, while 13q14-q21 loss and 5p14-pter gain has been associated with poorer outcome (Wang, Lu et al. 2001; Wang, Titley et al. 2003). The only anomaly occurring more than once in the study of 10 LMSs was amplification of 12q13-15, which includes genes of interest such as MDM2 and CDK4. Derre et al (Derre, Lagace et al. 2001) performed comparative genomic hybridisation (CGH) on 27

~~37 The cytogenetic anomalies reported in sarcomas with complex karyotypes was introduced in Chapter One (Section 1.8.3.2) and summarised in Table 1.7~~

~~298~~

~~Chapter 7: LMS-LFS Cell Line~~

LMSs, noting that 19 of the tumours had losses involving chromosome 13, with the putative common region of loss being 13q14-21. The observed rearrangements in this LMS-LFS cell line were, however, distinct from those described previously in the literature. There was no amplification of 12q13-15 or loss of 13q14-21 in this cell line. Research to date has not found an association between tumour grade or the development of metastases to a particular rearrangement or aberration. The prognostic significance of the specific anomalies seen in this cell line remains to be detemined. The considerable chromosomal aberration seen in the cell line, however, correlates with the high grade of the tumour from which it was derived and is a marker for adverse outcome. Additional investigation using a combined approach with array comparative genomic hybridisation (array CGH), CGH and spectral karyotyping may further elucidate the role of cytogenetic aberrations in carcinogenesis and tumour progression (Weng, Wejde et al. 2004).

~~7.3.5 ALT and telomerase in LMS-LFS cell line~~

It is known that immortalisation and possession of a telomere maintenance mechanism (TMM) are associated. Immortalisation may not be absolutely necessary for oncogenesis, particularly in cases where vascular supply to the tumour cells is not a limiting factor, or where the number of cumulative mutations required for carcinogenesis is small, but in the majority of solid tumours, it is an important factor linked to tumourigenesis. Tumour growth is dependent on the presence of an adequate vascular supply and therefore, when the degree of neovascularisation is no longer able to sustain ongoing tumour growth, central necrosis due to hypoxia may occur. The tumour may then also fail to proliferate unless other mechanisms come into play to circumvent this limitation. The Hayflick limit (maximum number of population doublings) is thought to be consistent with the development of large tumours without the cells necessarily being immortal (Reddel 2000). Malignant transformation is also contingent on the accumulation of genetic changes and clonal expansion of the transformed cell. If the number of genetic changes required for a particular malignancy are few, fewer population doublings will be needed to reach this stage and immortalisation may again, not be a prerequisite.

~~299~~

~~Chapter 7: LMS-LFS Cell Line~~

Shay et al demonstrated that 85% of tumours utilise telomerase as their TMM (Shay and Bacchetti 1997) and maintain their telomeres at <10 kb. The catalytic subunit of telomerase is a reverse transcriptase TERT (Kilian, Bowtell et al. 1997), which has been used as an expression vector to immortalise cells (Yeager, Neumann et al. 1999). Other immortal cells have ALT (alternative mechanism for lengthening of telomeres), probably using processes that involve recombination. A DNA strand from one telomere of one ALT positive (ALT+ ) cell anneals to the complementary strand of another cell and uses it as a copy template, priming synthesis of new telomeric DNA (Reddel 2003).

ALT is characterised by the presence of ALT-associated PML bodies (APBs) and heterogeneous telomere lengths (Mean 20 kb; Range 3-50 kb) (Bryan, Englezou et al. 1995; Bryan, Englezou et al. 1997; Tsutsui, Kumakura et al. 2003). PML bodies are thought to be involved in cell cycle regulation, apoptosis, tumour suppression and possibly the repair of telomeric DNA recognised as being damaged. The LMS-LFS cell line expresses ALT-associated PML bodies.

The prevalence of ALT was examined in adult and paediatric soft tissue sarcomas (STS), (Henson, Hannay et al. 2005). 50% of high grade STS and 33% of metastatic STS were ALT+, with the highest prevalence being in malignant fibrous histiocytomas (MFH). The overall prevalence of ALT in STS appears to be higher than in epithelial tumours (Henson, Hannay et al. 2005). It has been suggested, that there may be greater repression of telomerase in mesenchymal cells with a relative greater probability of activating ALT. Mesenchymal cells have a slower turnover than epithelial cells and hence less telomere shortening. Immortal LFS cells have also tended to be ALT+, possibly due to loss of p53 function being an early event (Henson, Neumann et al. 2002). This is consistent with our telomerase negative, ALT+ cell line.

It is important to note that inhibition of telomerase has been known to cause senescence (Colgin, Wilkinson et al. 2000; Ludwig, Saretzki et al. 2001), while repression of ALT in an ALT+ cell line by fusion with a normal diploid fibroblast cell line or by endogenous expression of hTERT (Perrem, Bryan et al. 1999; Perrem, Colgin et al. 2001) has also resulted in cell death. Fusion of an ALT+ cell line with the telomerase

~~300~~

~~Chapter 7: LMS-LFS Cell Line~~

~~positive cell line HT1080, however, resulted in an immortal hybrid that utilised telomerase and repressed ALT.~~

The genes responsible for repression of ALT and indeed the genes involved in ALT itself, remain largely undefined. Elucidation of these genes and the involved downstream mechanisms will hold the key to the development of novel therapeutics directed against the TMM of immortal tumour cells.

~~7.4 SUMMARY and PERSPECTIVES~~

To our knowledge, this is the first cell line derived from a sarcoma in a patient with Li Fraumeni syndrome. This new cell line may therefore not only be useful for investigating the effects of p53 inactivation but also provide an invaluable in vitro model for investigating the relationship between germline p53 mutation and activation of ALT. Given that a large proportion of sarcomas appear to utilise ALT, the cell line may also serve as an in vitro model for TMM inhibitors. Other targeted therapeutics such as protein kinase C inhibitors could also be investigated using this cell line for preclinical studies.

~~301~~

~~Chapter 8: Summary and Conclusions~~

~~2"Z~~

~~8. GENERAL CONCLUSIONS and FUTURE PERSPECTIVES~~

~~302~~

~~Chapter 8: Summary and Conclusions~~

~~8.1 INTRODUCTION~~

Soft tissue sarcomas are tumours that affect a younger group of patients than other malignancies. They are increasing in incidence and are responsible, relative to their incidence, for a disproportionate number of deaths. Despite recent improvements in surgical techniques, neoadjuvant chemoradiation and adjuvant therapies, 50% of patients continue to die of their disease. The major reason for this is the development of metastatic disease, which is largely unresponsive to conventional chemotherapeutic agents. It is imperative therefore, to direct the attention of researchers to two main objectives. Firstly, to devise means of predicting which patients are at risk of developing metastases and secondly, to develop novel therapeutic means of treating those with, and those at risk of, developing metastases.

Over the last twenty five years, histologic tumour grade, out of all clinical prognostic indicators, has consistently been shown to be the strongest predictor of metastasis and mortality (Chapter 1, Section 1.4.2 and Table 1.4). We therefore considered the genetic factors associated with tumour grade to hold the key to understanding tumour progression and metastasis in soft tissue sarcomas.

In light of the above, the aims of this thesis were • to profile gene expression patterns associated with tumour progression and metastasis using three sarcoma cell lines of increasing metastatic potential, • to validate the expression of EGFR, which had been found to increase with tumour grade using real time reverse transcriptase polymerase chain reaction (RT-PCR), • to examine the protein expression of EGFR and its signal transducers on a large number of tumour samples using tissue microarray technology and finally, • to concurrently initiate, establish and characterise new sarcoma cell lines that can, in the future, serve as in vitro models for functional studies on EGFR and other prognostic markers of sarcoma tumour progression. This work utilised two high throughput techniques. Gene expression arrays were used to examine the expression of thousands of genes on individual samples, in this case, in

~~303~~

~~Chapter 8: Summary and Conclusions~~

three sarcoma cell lines. Real time RT-PCR enabled the quantitation of gene expression, evaluating multiple candidate genes in multiple samples, in order to validate the findings of the expression arrays. Tissue microarrays were used to examine the protein expression of candidate markers on clinical tumour samples.

~~8.2 Global Gene Expression Patterns and the Tumour Progression Model~~

Statistical analyses carried out on the results of the expression arrays examined genes differentially expressed between the low and high grade sarcoma cell lines, as well as across the low grade, high grade and metastatic cell lines. This revealed a large number of potential prognostic markers in a variety of functional groups.

~~Growth factors and growth factor receptors featured in the genes overexpressed in the high grade and metastatic cell lines, including members of the EGFR, TGFB, VEGF and IGF family.~~

~~8.2.1 EGFR~~

EGFR and upregulation of members of this pathway have been considered in detail in the various discussion sections of chapters 2, 3 and 4. EGFR over-expression and its correlation with tumour grade was examined and confirmed at the gene expression level on oligonucleotide arrays. While hundreds of genes are up- and down-regulated on gene profiling studies such as this, an important way to judge the significance of the findings is to analyse the pathways involved. EGFR was selected for further study as factors across all stages of its pathway, from ligand and receptor through to signal transduction were over-expressed in the high grade and metastatic cell lines. This indicated a possible role for this pathway in sarcoma tumour progression and metastasis and clearly presented interesting possibilities for both diagnostics, in terms of risk categorisation and therapeutics in the management of sarcomas.

~~8.2.2 TGFB~~

~~In contrast to EGFR, the conventional transducers of TGFB signaling, SMADs 1 to 4, were not over-expressed, although SMAD5 (MADH5), which shows strong homology to~~

~~304~~

~~Chapter 8: Summary and Conclusions~~

the other SMADs, was overexpressed. This raised the possibility of SMAD-independent pathways being active in TGFB signaling within sarcomas, such as through Cyclin B2 (CCNB2). CCNB2, which was overexpressed, is known to bind TGFB receptor II and hence play a role in TGFB-mediated cell cycle control (Liu, Wei et al. 1999). SMAD- independent signalling via the Ras-MAPK pathway has been suggested previously, as TGF-beta-activated kinase (TAK1 or MAP3K7) was upregulated in the high grade cell line38. Another key potential role for TGFB in sarcoma progression, as discussed in Chapter 4, is through transactivation of EGFR.

~~8.2.3 Other Candidate Genes~~

IGFBP2 and IGFBP7 were overexpressed in the high grade cell line. The IGF family was not examined further in this thesis, as earlier work conducted in our laboratory had already confirmed the predictive characteristics of IGFBP2 protein expression in STS (Busund, Ow et al. 2004). The findings of the gene expression arrays in this thesis were therefore in agreement with the above. IGFBP2 expression had, in addition, been examined by others in the context of synovial sarcomas (Allander, Illei et al. 2002).

The differential expression of ACVR2B and that of NDRG1 in sarcomas of different histologic grade are novel findings. NDRG1 (Chapter 3, Section 3.4.1.3) has been associated in a variety of epithelial cancers with both tumour progression and cellular differentiation. To our knowledge, this study is the first report of NDRG1 dysregulation in sarcoma. Little is known of its mechanism of action or indeed the pathways in which it may interact.

~~Integrin expression has previously been studied in the more common STS subtypes as well as sarcomas of the small round blue cell variety (SRBCTs) (Barth, Moller et al.~~

~~38 TAK1 (MAP3K7) which has a putative N-terminal protein kinase domain, regulates transcription by transforming growth factor-beta (Dempsey, Sakurai et al. 2000).~~

~~305~~

~~Chapter 8: Summary and Conclusions~~

1995; Benassi, Ragazzini et al. 1998). SRBTs and MFHs did not express ITGA2 protein but ITGA6 expression was shown in the former. Integrin expression is likely to be complex and subtype specific but its correlation with tumour grade and metastasis certainly warrants closer attention in terms of the potential for therapeutic agents to be directed against these markers. The over-expression of ITGA2 and ITGA6 in the high grade cell line was validated in this study using qRT-PCR.

Other candidate genes differentially expressed included those involved in the Notch pathway (NOTCH4 and DNER), Wnt pathway (DAAM1 and WNT5B), Ras-MAPK signal transduction pathway (MAP4K4, MAP3K7, MAPK13 and MINK1), cell motility and invasion (MMP1, MMP3, MMP14, LAMB3 and LAMR1), cell cycle regulation (CCND1, CCNB2, CDC25C, CDKN1A, BID and BNIP1) as well as numerous protein/tyrosine kinases, their substrates and receptors (PRKCD, PRKCBP1, TIE, SRPK1, to name a few). Many of these were discussed in chapter 2 (Sections 2.4.4.3.3 to 2.4.4.3.6). Time constraints would clearly not allow all of these to be evaluated in greater detail, however, they will serve as starting points for future studies on sarcoma progression.

Additionally, what can be speculated from the global gene expression analysis in this project, is the possibility of considerable cross talk between pathways, such as that between EGFR and TGFB and matrix metalloproteases, integrins and cell cycle regulators. Intermediaries in these signaling processes are the MAPK/Erk, PKC and PI3k/Akt pathways. Cross talk between G-protein coupled receptors (GPCRs) and EGFR, or EGFR transactivation, was discussed in chapter 4 (Section 4.4.3). GPCR mediated transactivation is carried out via members of the ADAM (a disintegrin and metalloprotease) family of proteases (Fischer, Hart et al. 2006), which in turn have been implicated in Notch signalling (Blobel 2000). Transactivation of EGFR by TGFB has also been noted (Caja, Ortiz et al. 2007) in other epithelial malignancies and could potentially be at play in sarcomas.

~~EGFR has been the subject of much recent interest and investigation in many cancers (Chapter 2, Section 2.4.4.3.2 and Chapter 3, Section 3.4.2.2.6), particularly in relation to~~

~~306~~

~~Chapter 8: Summary and Conclusions~~

~~therapeutic EGFR inhibition. Its expression in sarcomas of different grade was examined in greater depth in this thesis with tissue microarray.~~

~~8.3 EGFR Activation and Signal Transduction in Sarcoma~~

It is known that gene or mRNA transcript expression does not necessarily correlate directly with protein expression (Chuaqui, Bonner et al. 2002). In other words the genome, transcriptome and proteome do not precisely correlate. Other influences are at play, such as epigenetic phenomena and post translational modification. Thus, while gene expression arrays greatly expedite the process of gene discovery, the expression of the protein products of these genes must be verified.

In this thesis, tissue microarray was employed as a tool for examining EGFR protein expression in soft tissue sarcomas of varying histologic subtype and grade. We believe this is the first body of work to examine EGFR function in the context of tumour progression in sarcoma. This was achieved by evaluating the expression of the activated form of EGFR, and that of three of its signal transducers across STS of varying grades, including metastases.

As noted in previous chapters, this transmembrane tyrosine kinase receptor mediates cell proliferation, survival, adhesion, migration, angiogenesis and differentiation. Its over-expression has been reported in epithelial cancers as well as certain histologic subtypes of sarcoma, particularly synovial sarcoma, malignant peripheral nerve sheath tumours and paediatric-type rhabdomyosarcomas (Section 2.4.4.3.2). Recently, more widespread EGFR overexpression among other common histologic subtypes of sarcoma has been reported (Section 4.4.3) (Sato, Wada et al. 2005). For example, this Japanese study supported ours in that wild-type EGFR expression did not predict survival on multivariate analysis. The anti EGFR antibody used in the Japanese study was from a different source (EGFR PharmDx kit, DakoCytomation, Glostrup, Denmark) and specific for the extracellular ligand binding domain of the EGF receptor. However, the expression of phosphorylated EGFR was not examined in their whole patient cohort.

~~Our study was able to confirm the correlation between tumour grade and the expression of EGFR, activated (phosphorylated) EGFR and the phosphorylated (activated) forms of~~

~~307~~

~~Chapter 8: Summary and Conclusions~~

44/42MAPK (Erk1/Erk2) and Akt. Importantly, a novel finding in this thesis was that activated EGFR and pAkt were independent predictors of decreased survival. These findings not only highlight the prevalence of EGFR overexpression in sarcomas of all the common histologic subtypes but also indicate that it is present in its activated form. That PI3K-Akt and Ras-MAPK pathways may both be involved in EGFR signal transduction within sarcomas, was again a novel finding.

These results hold promise for targeted therapeutics, despite the disappointing results of EGFR inhibition in non small cell lung cancer (NSCLC), where only those with activating EGFR mutations in the tyrosine kinase domain were found to respond (Lynch, Bell et al. 2004; Paez, Janne et al. 2004). It is not known whether mutations in the EGFR receptor will be a necessary component of responsiveness to EGFR-directed treatment in sarcoma. Indeed, different mechanisms have been shown to confer sensitivity or resistance in other malignancies (Thomas, Chouinard et al. 2003; Haas- Kogan, Prados et al. 2005; Yang, Qu et al. 2007).

~~8.4 Initiation and Characterisation of Sarcoma Cell Lines~~

One of the problems faced in the field of sarcoma research is the dearth of sarcoma cell lines available that could potentially serve as in vitro models for preclinical studies. Cell lines exist for the rarer sarcomas of simple karyotype, such as Ewing’s sarcoma and synovial sarcoma, whereas there are few reports of cell lines derived from sarcomas of complex karyotype such as leiomyosarcomas and pleomorphic sarcomas (Teicher 1999). Of the sarcomas of simple karyotype, there are only two reports of GIST cell lines in the literature and a few short term cultures (Tuveson, Willis et al. 2001; Taguchi, Sonobe et al. 2002; Duensing, Medeiros et al. 2004).

Thus, an additional aim of this thesis was to initiate primary cultures from fresh sarcoma tissue and where these were established as cell lines, to characterise them in terms of morphology, biomarker expression, cytogenetics, ALT and telomerase activity. A number of cell lines survived repeated passaging and two of these, a GIST cell line (GIST-M) and a cell line derived from a leiomyosarcoma (LMS-LFS) were characterised fully.

~~308~~

~~Chapter 8: Summary and Conclusions~~

The GIST-M cell line possessed the same KIT mutation as the original tumour, demonstrated cytogenetic characteristics that typify GISTs, such as loss of chromosomes 14 and 22. As the original tumour was resected from the patient after she had received six months of Imatinib treatment, with little change in symptoms or tumour size, it was postulated that the cell line may carry a secondary mutation. However, this proved not to be the case. GIST-M was found to over-express phosphodiesterase 2A (PDE2A). This marker is thought to be associated or coregulated with KIT, showing downregulation with Imatinib treatment in other studies. The cell line could therefore be used for in vitro studies of the effects of Imatinib treatment, or that of other therapies.

The leiomyosarcoma from which the LMS-LFS cell line was derived arose in a patient with Li Fraumeni syndrome and to our knowledge, there are no other reports of cell lines established from a sarcoma in a Li Fraumeni patient. This cell line therefore possesses the same germline TP53 mutation present in the patient and the original tumour. In addition, it was found to demonstrate ALT positivity, which is clearly its means of escaping senescence. It was also shown to overxpress protein kinase C theta (PRKCQ), which has not previously been reported in leiomyosarcomas. This cell line therefore provides the means for investigating the effects of p53 mutation, ALT in a sarcoma cell line, as well as the role PRKCQ plays in this tumour. ALT and PRKCQ are potential therapeutic targets.

~~8.5 Summary~~

• Gene expression arrays and tissue microarrays are complementary tools that can be effectively utilised in soft tissue sarcoma research for gene discovery and validation on clinical samples. • A large number of differentially expressed genes were identified, yielding potential prognostic markers requiring further investigation in clinical samples. • The differential expression of EGFR and other selected genes was validated using real time RT-PCR. • Tissue microarray enabled cellular and to some extent, subcellular localisation of the biomarkers of interest in soft tissue sarcoma samples of different histologic grade.

~~309~~

~~Chapter 8: Summary and Conclusions~~

• Clinical correlation and cox regression analysis proved activated EGFR and phosphorylated Akt to be independent predictors of survival in our group of patients. In addition, phosphorylated MAPK expression correlated with poor survival on univariate analysis. The results suggest that these biomarkers could be the subject of targeted therapeutics and further functional studies. Moreover, they could be used for prognostication. • Primary cultures of a variety of subtypes of sarcoma were initiated and two of the cell lines which survived repeated subculturing were fully characterised. It was not within the scope of this thesis, owing to time constraints, to fully characterise several other cell lines that also survived repeated passaging. These cell lines were established in response to the perceived need to add to the sarcoma cell line repository for future research.

~~8.6 Future Directions~~

~~8.6.1 Bioinformatics~~

The gene expression data in this study was analysed in a number of ways. Both GeneSpring v7.0 software (Silicon Genetics, Agilent Technologies, Palo Alto, CA) and Bioconductor R statistical package were used independently to normalise and analyse the data. Clustering methods were used within GeneSpring to aid visualisation of co- expressed genes. The statistically significant genes and gene clusters were subsequently examined further using the web based annotation tool Database for Annotation, Visualisation and Integrated discovery (DAVID) (http://apps1.niaid.nih.gov/david/). The DAVID Knowledgebase interrogates and integrates dozens of public databases comprising >40 million records for gene ontology, protein domains, disease associations, functional categories and pathways. The algorithms used utilise gene-annotation enrichment analysis, as well as functional annotation tools such as the DAVID Gene Functional Classification Tool and DAVID Functional Annotation Clustering Tool for additional biological interpretation. Links to the DAVID Pathway Viewer allows examination of genes of interest dynamically on pathway pictures, such as BioCarta Pathways and KEGG Pathways, as illustrated in the thesis in Figures 2.5, 2.7, 2.9, II.1 (Gene Ontologies) and 2.12, 2.13, II.2, II.3 (KEGG pathways). The DRAGON database (Database Referencing of Array Genes Online)

~~310~~

~~Chapter 8: Summary and Conclusions~~

~~(http://pevsnerlab.kennedykrieger.org/dragon.htm), another freely available web-based annotation platform was also interrogated for the genes of interest in this thesis.~~

Other bioinformatic annotation platforms such as Gene Set Enrichment Analyses, Connectivity Platform and Ingenuity utilise similar algorithms for ontology annotation and visualisation of pathways. These could have been additionally used. However, it was felt that it would be unlikely that these would provide additional data of biologic significance. Ultimately, the experimental design of this study was such that the microarray data was used to generate hypotheses that were subsequently examined and validated at protein expression level.

Supervised learning methods are used in certain gene expression array studies, requiring two sets of samples. The first is a “learning set” and the second, a “testing” or “validation” set. These are then used for the purpose of class prediction, among others. As only 3 cell lines were used in this gene expression study, this experimental design would not have been appropriate to use in this setting.

~~8.6.2 Functional Studies~~

Based on the findings of this thesis, preclinical trials of EGFR inhibition in sarcoma cell lines have commenced in our laboratory. These could, in addition, be used to examine the mechanism of action of EGFR, in particular the downstream signal transducers. Gene expression changes with drug treatment could also be assessed on a global scale using gene expression arrays. This may help to fully elucidate the interaction between pathways in sarcoma tumourigenesis and progression.

~~8.6.3 Upstream of EGFR~~

The signal transduction pathways downstream of EGFR were examined in the present study, presenting some novel data. Examining the mechanism of activation of EGFR itself presents another avenue of future study. Ligand dependant and independent mechanisms are known to play a role in EGFR activation and transactivation and some of these have been discussed in this thesis (Chapter 4, Section 4.4.3). While these mechanisms are known to exist in other normal and malignant tissues, their role in sarcomas in the context of EGFR activation requires further exploration. Epiregulin, a

~~311~~

~~Chapter 8: Summary and Conclusions~~

~~ligand of EGFR was upregulated in the higher grade sarcoma cell lines on the gene expression arrays, as was TGFB1, which can transactivate EGFR.~~

~~8.6.4 Targeted Therapy in STS~~

As alluded to in previous chapters, lessons have been learnt from the use of EGFR targeted therapies in other malignancy and KIT-directed therapy for GISTs. In the latter, the emergence of secondary mutations has conferred resistance to Imatinib, and most recently, to Sunitinib (Antonescu, Besmer et al. 2005; Debiec-Rychter, Sciot et al. 2006; Demetri, van Oosterom et al. 2006). A subset of glioma patients expressing a variant of the EGF receptor are also resistant to EGFR inhibition (Haas-Kogan, Prados et al. 2005), while some studies had noted that activating mutations of EGFR were required for response to Gefitinib in NSCLC patients (Lynch, Bell et al. 2004; Paez, Janne et al. 2004). A large randomised placebo controlled trial of 731 patients with Stage IIIB or IV NSCLC using Erlotinib, however, reported different results (Shepherd, Rodrigues Pereira et al. 2005; Tsao, Sakurada et al. 2005a). The tumours from 325 of these patients were analysed for EGFR expression by immunostaining, 177 had mutational analyses performed and gene amplification was examined in 221. In this case, the presence of EGFR mutation did not correlate with EGFR expression, gene amplification, responsiveness to Erlotinib or survival. Patients treated with Erlotinib who were found to have amplification of EGFR did have a survival benefit over those on placebo. Further analysis correlating progression free survival with presence of EGFR expression, mutation or amplification was not reported. The expression of activated EGFR was not examined.

A recent study of STS revealed no EGFR mutations in the cohort examined (Baird, Davis et al. 2005). However, as discussed in Chapter 4 and above, EGFR mutations are not the sole determinants of responsiveness to EGFR-directed therapy. These factors as well as activated EGFR and copy number changes need to be studied in order to determine which factors and mechanisms will be necessary components of responsiveness to EGFR-directed treatment in sarcoma. The complexity of the dysregulated signals within all malignancies indicates that multiple factors are likely to be driving tumour progression and metastases. A

~~312~~

~~Chapter 8: Summary and Conclusions~~

combination of conventional chemotherapeutic agents produces greater results in treating advanced malignancies by targeting different aspects of cell biology. Thus it is likely that these newer targeted agents will also need to be used in conjunction with other agents in a “multi-pronged” approach in order to achieve the best results. If the drug combinations can be tailored to the risk profiles of individual patients, as determined by the profile of prognostic biomarkers expressed by their tumours, then the therapy will truly be “targeted”. Many newer agents have been licenced for use only in advanced malignancies, where other chemotherapeutic agents have failed to induce a response. However, the patients may in fact be better served by receiving these treatments early in the course of their disease, acting in synergistic fashion with other so-called first line agents. The challenge of targeted therapy in STS has been taken up both in terms of combination therapy and in terms of broader spectrum agents targeting, for instance, more than one tyrosine kinase (Okuno 2006; Hartmann 2007; Kasper, Gil et al. 2007; von Mehren 2007). Perhaps the moderate success of these targeted treatments highlights the complexity of the metastatic process. There remains a pressing need therefore to continue to further our understanding of sarcoma tumour biology and identify more prognostic markers, in the effort to successfully manage our patients who present with this disease.

~~313~~

~~Chapter 8: Summary and Conclusions~~

~~314~~

~~Appendices~~

~~"() 9~~

~~I. PROTOCOLS and ADDITIONAL DATA for MICROARRAY and CELL CULTURE EXPERIMENTS~~

~~315~~

~~Appendices~~

~~I.1 Design Issues in Expression Profiling~~

Gene expression profiling is a technique that has now generated sufficient experience within the research community to be able to grasp the importance of experimental design and a number of reviews have dealt with these issues (Churchill 2002; Dobbin and Simon 2002; Yang and Speed 2002; Dobbin, Shih et al. 2003; Holzman and Kolker 2004). There are both scientific and logistic issues that govern the process of decision making when considering study design. The former deals with ascertaining whether the design of the experiment will answer the questions posed. The latter is concerned with the practical aspects such as availability of sample material, choice of reference RNA and standardisation of all protocols to control for experimental variability.

~~I.1.1 Replicates~~

The number of biologic and/or technical replicates has to be determined. Biological replicates refer to different RNA extractions (RNA from different patients or RNA from a cell line at different passages/subcultures) and technical replicates refer to repeat hybridisations carried out using RNA from the same extraction. Replication is essential in experimental design because it allows an estimate of variability.

~~I.1.2 Experimental Design~~

The design of the experiment may be direct, where two samples of interest are compared, or indirect, where many samples are compared to a common reference (Dobbin and Simon 2002; Yang and Speed 2002). Variants of the direct design are loop designs (Kerr and Churchill 2001a; Kerr 2003), where samples are successively compared. The inherent problems with the loop designs are the complexity of the statistical analysis that has to be performed and difficulty in determining the accuracy of the comparisons if the samples are far apart on a large loop.

The simple common reference design was chosen and the reference cell line was always labelled with green fluorescent dye (Kerr and Churchill 2001a; Churchill 2002; Yang and Speed 2002; Dobbin, Shih et al. 2003). Loop designs are not “robust”, in that the designs may “become disconnected” if there is loss of data from arrays. The common

~~316~~

~~Appendices~~

reference design leaves other hybridisations untouched even if some arrays are lost or have data that cannot be used and hence is the more robust design. Pooled RNA from the normal fibroblast cell line was chosen as the reference, as it could be cultured and harvested in identical manner to the sarcoma cell lines. It could be cultured in sufficient quantities such that aliquots from the same pooled sample could be used on all experiments, thus reducing variability. It also gave a signal in most spots, which is important as the ratio of the signal is what is used as the indirect measure of the intensity in the sarcoma cell lines. Study designs that compare “diseased” to “normal” may find the latter difficult to define (King and Sinha 2001). “Normal” cannot often be standardised for all patients, particularly with some types of STS, where the normal cell counterpart is not known39. Tissue taken adjacent to a tumour may also not be “normal”. In cases where clear surgical margins cannot be obtained, it may not be possible to obtain adjacent normal tissue. This issue is circumvented in my study by using the common reference design, where different tumour cell lines are being compared and the choice of the normal fibroblast cell line MRC5 is in some ways immaterial, in that it is not being used as an example of “normal”, it is simply a reference, as in other studies where a pool of different cell lines is used as the reference.

~~I.1.3 Platform: Oligonucleotide arrays~~

Two main available options include the in situ synthesised oligonucleotide arrays, such as the commercially available Affymetrix arrays or spotted cDNA or oligonucleotide arrays. The spotted arrays can be made in-house by academic institutions. cDNA arrays and oligonucleotide arrays differ in the number of bases that are printed for each gene. cDNA arrays are 500-2000 bases long, which means more than one PCR fragment may end up as a probe during its synthesis, and in some cases be absent, which can be a

39 This concept was discussed in Chapter 1, Section 1.4.2, where degree of differentiation, one of the crieteria for histologic grading if STS was described. The cell of origin of STS subtypes such as clear cell sarcoma, alveolar rhabdomyosarcoma and epithelioid sarcoma.

~~317~~

~~Appendices~~

problem. However, differential splicing is generally less of an issue. Oligonucleotide arrays are 20-30 bases in length for the in situ synthesised Affymetrix arrays and 70-mer for custom spotted arrays printed in-house. 70-mer is considered the optimal length to limit cross hybridisation. For these reasons, and the expense involved in purchasing arrays commercially, the Human 19K oligonucleotide arrays printed by the Ramaciotti Centre, UNSW and the Adelaide Microarray Facility were used for this study. For initial optimisation experiments, slides from the Ramaciotti centre were used. In later experiments, due to timing of print runs and availability of slides, the remainder of the slides was obtained from the Adelaide Microarray Facility. Both institutions printed the oligonucleotide library from Compugen (Compugen 19,000 human oligo library Release 1) manufactured by Sigma-Genosys. The validity of using microarrays for gene expression analysis was called into question when a number of cross platform comparisons showed poor correlation of the data obtained from studies by different research groups using similar patient samples or even the same cell lines (Kothapalli, Yoder et al. 2002; Kuo, Jenssen et al. 2002; Gabor Miklos and Maleszka 2004). These groups typically compared cDNA microarrays to Affymetrix in situ synthesised oligonucleotide arrays. Lack of probe specificity, with cross hybridisations and incorrect sequence verification were cited as causes of the poor correlation between platforms. 3′ bias within all expression arrays over this time presented additional difficulties. Other studies have, however, shown concordant results in their cross platform comparisons (Rhodes, Barrette et al. 2002; Sorlie, Tibshirani et al. 2003; Jarvinen, Hautaniemi et al. 2004; Larkin, Frank et al. 2005). Jarvinen et al performed hybridisations using the same breast cancer cell lines on four different array platforms. The correlation coefficients ranged from 0.62 to 0.86, with the two commercial platforms exhibiting greater correlation. These findings highlight some of the problems inherent in attempting to compare datasets and draw firm conclusions on the basis of a single gene expression study. Microarray protocols should be used as a powerful tool capable of generating hundreds of hypotheses that then require validation.

~~I.1.4 Dye Bias~~

~~Dye bias has to be taken into consideration. As dual colour arrays require the hybridisation of red and green labelled cDNA samples with the oligonucleotide probes~~

~~318~~

~~Appendices~~

on the slides, the ratio of the green to red fluorescence signal is what is obtained on scanning. This is therefore a measure of the relative expression levels of the transcripts from the two samples. As gene specific dye bias is known to occur, as well as differences in the dye incorporation efficiencies across all genes, these may confound the relative gene expression patterns observed. Normalisation of the data (see below) will correct for the dye incorporation differences but individual gene-specific dye binding differences will not. This phenomenon has led to some researchers performing dye swap experiments for the samples being compared to control for this confounder. However, dye swaps are not mandated for all experimental designs. If the two samples are being directly compared, for instance where there is paired tumour and adjacent normal tissue taken from the same patient, or pre and post treatment samples from each patient, dye swap experiments should be carried out. Similarly, if the sole aim is to compare just two samples, such as a tumorigenic and non-tumorigenic cell line, dye swaps are required. If several samples are to be compared across multiple arrays, and an indirect design is used with a common reference, dye swaps only need to be performed if comparison to the reference is also of interest.

~~A common reference design can be used without dye swaps where the comparison of interest is between samples that are all labelled with the same fluorescent dye.~~

~~I.1.5 Normalisation and Analysis of Data~~

~~I.1.5.1 Normalisation~~

There are numerous sources of systemic variation in microarray experiments, including, as alluded to in the previous section on dye bias, the differing labelling efficiencies of the fluorescent dyes. Dye bias is a function of signal intensity. Dye bias can also vary with spatial position on the slide, due to differences between the print-tips on the array printer, variation over the course of the print-run, among others. Differences between arrays may arise from differences in print quality or even the ambient temperature and humidity at which the plates were processed. Other sources include everything from differences in RNA quality to post hybridisation stringency washes and scanning properties. The effects of these variations can be large relative to the actual biological differences between the samples.

~~319~~

~~Appendices~~

The purpose of normalisation is to remove systemic sources of variation such that the real biologic differences between samples can be measured (Quackenbush 2002; Yang, Dudoit et al. 2002; Bolstad, Irizarry et al. 2003; Smyth and Speed 2003). For the reasons outlined above, normalisation needs to be intensity dependent and carried out between as well as within arrays. Global per spot per chip intensity dependent (LOWESS)40 normalisation the most commonly used method for achieving this aim. Local regression estimation is a method for smoothing scatterplots. The mathematical models for this are described by Yang and colleagues and in a review by John Quackenbush (Quackenbush 2002; Yang, Dudoit et al. 2002).

~~I.1.5.2 Data Analysis~~

Numerous methods have been employed in microarray data analysis. Most early studies used “fold change” above a user defined threshold to present the data. Common current practice applies a variety of standard statistical tools modified for use in this setting where large datasets require analysis. This includes parametric tests such as the t-test and for comparison across more than two groups, the analysis of variance, or ANOVA41 (Kerr, Martin et al. 2000; Slonim 2002). Replicates are essential with these statistical tests. In addition, multiple testing correction is applied to reduce the false discovery rate (FDR). This adjusts the individual p-value for each gene to keep the overall error rate or FDR to within the user-defined cutoff. Standard methods such as Bonferroni, Westfall and Young permutation or Benjamini Hochberg (decreasing order of stringency) can be

40 LOWESS refers to Localised Weighted Regression. This was originally described in Cleveland, W. S. and S. J. Devlin (1988). Locally-weighted regression: An approach to regression analysis by local fitting. J. Am. Stat. Assoc. 83, 596–610.

~~41 ANOVA is used to assess the variability, or spread of data. Variance is the average squared deviation from the mean. It measures the spread of data around the mean.~~

~~320~~

~~Appendices~~

~~used for this purpose42. Higher order analyses such as clustering and class prediction can be carried out with the reduced list of statististically significant genes.~~

~~I.1.5.3 Clustering analysis~~

In this study, hierarchical clustering was used as a means of identifying potentially coregulated genes, with the proviso that similarity of gene expression does not necessarily indicate similarity of function. This revealed groups of genes that were over- or under-expressed in each of the sarcoma cell lines. A larger sample size than the one in the present thesis would have allowed the use of supervised clustering methods used for class prediction. However, Michiels et al delivered a blistering attack when they reanalysed the data from eight large studies on cancer prognosis (Michiels, Koscielny et al. 2005). Each of the reviewed studies had at least 60 patients. The molecular signature varied considerably with the selection of the patients that formed the training set. In their words, “the list of genes identified as predictors of prognosis was highly unstable”. Misclassification rates ranged from 31% to 49%. The results are sobering but the fact remains that gene expression studies have uncovered valuable gene associations. What this highlights, is the need to validate the results of gene expression studies using other complementary techniques.

~~I.1.5.4 Comparison of GeneSpring and Bioconductor R statistical package~~

Independent biocomputational techniques were used to analyse the gene expression data and ensure statistical significance, yielding complementary information. Different algorithms, including methods of ranking and post hoc corrections, are applied by the two programs to examine differentially expressed genes. Hence, although there is considerable overlap in the lists of differentially expressed genes generated by these

42 The listed methods for multiple testing correction can be applied to the ANOVA within the GeneSpring analysis program used for the microarray analysis in this study, as discussed in Section 2.2.8.1 under Materials and Methods. The Benjamini Hochberg false discovery rate was used in this case (Benjamini and Hochberg 1995). 321

~~Appendices~~

methods, there are some differences. This represents a further source of variation in microarray experiments. In the subsequent chapter on validation studies, the candidate genes are selected from both of these statistical analyses. Some of the selected genes are common to both sets of analyses. Significant proportions of the classified genes could be grouped into those involved in signal transduction pathways, inhibition of apoptosis, cell proliferation and degradation of extracellular matrix.

~~I.2 MTT ASSAY FOR CELL PROLIFERATION~~

All cell lines were harvested with 0.25% trypsin (GIBCO,Invitrogen, Carlsbad, CA) and plated into 10 96-well plates at initial densities of 1 × 103 cells/well, 2 × 103 cells/well and 5 × 103 cells/well in 200 μl of culture medium. This was done in triplicate for each cell line. The plates were incubated in standard culture conditions and media changed every three to four days. Daily for ten days, one plate was assessed for proliferation using the MTT assay according to the manufacturer’s protocol, with some modification. Media was removed and 200 μl of MTT at 1mg/ml in serum free media was added to each seeded well as well as 3 blank wells as a control. The plate was then incubated at

37°C and 5% CO2 for 4 h, after which, the MTT was removed and replaced with 200 μl dimethyl sulphoxide (DMSO) to dissolve the formazin crystals. Absorbance of samples at 450 nm was quantified using a Sunrise TouchScreen plate reader (Tecan, Salzburg, Austria) with Magellan software (Tecan).

~~I.3 QUALITY CONTROL OF PRINTING OF OLIGONUCLEOTIDE ARRAYS~~

These arrays comprised a commercial oligonucleotide library from Compugen (Compugen 19,000 human oligo library Release 1) and manufactured by Sigma- Genosys. The oligonucleotide sequences were optimised to account for a maximal number of splice variants for each given gene (using conserved regions of the gene for selection). Other aims of oligonucleotide selection and synthesis included minimisation of cross hybridisation, cross homology against other sequences on the array (average cross homology 30.8%) and secondary structure, as well as ensuring optimal GC content (average 50%) and optimal sequence length for binding to the substrate. The microarray slides were printed at the Clive and Vera Ramaciotti Centre for Gene Function Analysis, University of New South Wales

~~322~~

~~Appendices~~

(http://www.ramaciotti.unsw.edu.au) and the Adelaide Microarray Facility, University of Adelaide (http://www.microarray.adelaide.edu.au). Adelaide’s oligonucleotide library was purchased with Australian Cancer Research Foundation (ACRF) funds.

All data from the quality control hybridisations was normalised. The average intensity of the signal (A) for each spot was plotted against the log2 ratio of the red / green signal (M) (Figure 2.2). The slides were compared to each other on MvM plots (Figure 1.1), providing an assessment of differential gene expression across the slides. As the same cell lines labelled with the same fluorescent dyes are used for all quality control hybridisations, there should be minimal differential expression, provided the printing of the oligonucleotide probes has been uniform throughout the print run.

~~323~~

~~Appendices~~

Table I.1 Human 19K Control Spots < +## #1!0'.2'-, .-*78 .-*78 <82-+-,'2-0,-,V1.#!'$'! ',"',%-$.-*73! <8T 3$$#0 .-22',%1-*32'-,2-+-,'2-0-*'%-! 007-4#0T SVSZ SZb&-31#)##.',%c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

~~324~~

~~Appendices~~

~~ *'"# WR ," *'"# SRR *'"# S ," *'"# SRR~~

~~! *'"# WR " *'"#SRR~~

Figure I.1 Quality control (QC) hybridisations at Adelaide Microarray Facility The cell line MCF-7 is labelled with Cy3 and the cell line Jurkat, Cy5 for all slides. Slides from the beginning, end and middle of the print run are taken for the quality control experiments. The figures (a) and (b) show MvM plots for the Compugen Human 19K print run 9, where M refers to the log2 ratio of red to green signal and MvM is therefore a plot of the M value for the two arrays (slides) being compared. Each spot represents the expression ratio for one gene. As the same cell lines labelled with the same dye are used on all slides, the ideal MvM plot should be a straight line through the origin. The deviation of spots from this line provides a measure of how similar the QC arrays are within and between the print runs. In (c) and (d), the log average normalised fluorescent intensity of each spot, A (log2A(Red x Green)) for each gene is plotted against M. The "shape" of this MvA plot indicates the number and intensity of the signals. It also provides a measure of the differential expression on an array. Additional Bayesian statistics are used to assess the likelihood that genes are consistently differentially expressed across all QC arrays.

~~325~~

~~Appendices~~

~~I.4 RNA ISOLATION~~

Culture adherent cell line in 150 cm2 dishes Harvest when at 70-80 % confluence Add the appropriate amount of TRIzol directly to the dish (12ml) Scrape off cells with sterile scraper Transfer cell suspension into 1.5 ml eppendorf tubes and homogenise 5 min Add 0.2 ml chloroform per ml TRIzol and shake vigorously for 15 s Incubate tubes at RT for 15 min Centrifuge tubes at 12 000g at 4° C for 15 min After centrifugation, cell suspension should be divided into 3 layers: the bottom pink layer, middle white interphase and top clear aqueous layer. Transfer top aqueous layer into a fresh eppendorf tube Add 0.5 ml of isopropanolol (per ml of TRIzol) and mix by inverting the tube 5x Incubate for 10 min at RT to recover the RNA by precipitation Centrifuge at 12 000g at 4° C for 15 min Remove all isopropanolol, ensuring that pellet is not disturbed Add 1 ml 75% ethanol (per ml of TRIzol) and resuspend the pellet Centrifuge at 7500g at 4° C for 5 min Carefully remove all traces of ethanol Allow pellet to air dry for ~ 1 min, ensuring pellet does not disappear Resuspend pellet in 30-50 )l of RNase-free water Store the RNA sample immediately in the – 80° C freezer Check RNA quality (260 / 280 ratio) on GeneQuant using 3 μl sample and 147 )l of RNase-free water If required, RNA can be further purified with the RNeasy purification kit (Qiagen) before storing Check RNA quality and quantity on Agilent BioAnalyzer and NanoDrop prior to cDNA synthesis for microarray

~~326~~

~~Appendices~~

~~I.5 cDNA SYNTHESIS AND LABELLING~~

Slide No. ______Print Run ______Samples Sarcoma Cell Line ______BioAnalyzer (Date) Sample ___ NanoDrop Ratio BioAnalyzer Ratio Concentration Volume used Amt RNA used 20 μg Pooled MRC5 BioAnalyzer (Date) Sample ___ NanoDrop Ratio BioAnalyzer Ratio Concentration Volume used Amt RNA used 20 μg # 12#0#'6$-0T0# !2'-,1+ "#3..* !#"-, !#

~~5× 1st Strand Buffer 6 x 2 = 12μl DTT 1.5 x 2 = 3μl dNTP mix 1.5 x 2 = 3μl RNaseOUT 1 x 2 = 2μl SuperScript III 2 x 2 = 4μl ##2&-"SS12120 ,"!)(" 7,2'1~~

~~Oligo dT per tube RNA 2 μl Incubate @70oC 5min Place on ice 1min To above tubes, add Master Mix 11.7 μl~~

~~To each tube, add DEPC H2O upto 30 μl total volume~~

~~327~~

~~Appendices~~

Incubate @ 46oC 2 h To each tube, add 1N NaOH 15 μl Incubate @ 70oC 10 min To each tube add 1N HCl 15 μl To each tube, add Na+ acetate 20 μl ##2&-"S30'$'! 2'-,S12120 ,"!)("31',% ,4'20-%#, ("!-*3+,1

Loading Buffer per tube 500 μl Apply sample to SNAP column on collection tube Centrifuge @ 14000g RT 1 min Discard flow through Wash Buffer per tube 700 μl Centrifuge @ 14000g RT 1 min Discard flow through Repeat Wash Buffer 700 μl Centrifuge @ 14000g RT 1 min Discard flow through & centrifuge again 1 min Place SNAP column on fresh 1.5 ml tube

~~To each column, add DEPC H2O 50 μl Incubate @ RT 1min, then centrifuge 1 min Repeat step 12-13 Eluate contains cDNA in 100 μl ##2&-"S2& ,-*0#!'.'2 2'-,~~

Sodium acetate per tube 10 μl Glycogen per tube & mix 2 μl Ice cold 100 % ETOH per tube 300 μl Incubate tubes @-20° C for at least 30 min Centrifuge @ 4° C 14000g for 20 min Discard supernatant, ensure pellet remains Ice cold 75 % ETOH per tube 250μl Centrifuge @ 4° C for 2 min

~~328~~

~~Appendices~~

~~Discard supernatant Air dry samples, tubes inverted on tissue 10 min Resuspend pellet in 2X Coupling Buffer per tube 5 μl ##2&-"S* #**',%5'2&*3-0#1!#,2)7# ,) 0)~~

Previously aliquoted dyes resuspended with DMSO 5 μl Appropriate dye added to each tube (Cy3 or Cy 5) Mix well & incubate in dark @ RT at least 1 h Short Blocking protocol for slide during this period ##2&-"S &-02&*-!)',%0-2-!-*$-0-,#S['#'!0- 00 7 *'"# .0'-02-&7 0'"'1 2'-,Q.-12V.0-!#11',%-$2*'"#'1! 00'#"-322-!&#+'! **7 !-,4#022&#.-1'2'4#.-*7*71',#130$ !#2-.0#4#,2,-,V1.#!'$'!&7 0'"'1 2'-, To make 0.1 % SDS 50 ml, add 10 % SDS stock 0.5 ml

~~To ddH2O 49.5 ml To make 5% ETOH 50 ml, add 95%ETOH 2.63 ml~~

~~To ddH2O 47.5 ml rd 3 50ml tube contains ddH20 50 ml Microwave 0.1% SDS tube with cap off for 25 s Allow to come to 95° C Wash slide in 0.1 % SDS, constant stirring for 1 min~~

~~Wash in 5 % ETOH solution in ddH20 @ RT 1 min~~

Wash in ddH20 @ RT constant stirring @ RT 1 min Dry slide by spinning @ 2000 rpm for a few seconds. Keep on bench with tissue covering, ready for hybridisation ##2&-"S30'$7',%* #**#"!)("31',%? "/3'!)0-2-!-* Add 100 % ETOH to Buffer PE before use Add Buffer PB per tube and mix 50 μl Place QIAquick column on 2ml collection tube Apply sample to column & spin @ 18000g for 1 min Discard flow through Add Buffer PE to QIAquick column 750 μl

~~329~~

~~Appendices~~

~~Centrifuge @ 18000g RT for 1 min Discard flow through Centrifuge for an additional 1 min Place QIAquick column on fresh 1.5 ml tube~~

To elute, add DEPC H2O to column 30 μl Let the column stand for 1 min Centrifuge @ 18000g for 1 min. This results in 30 μl purified labelled cDNA Speed Vac dry @ 45° C down to 2μl in approx 20 min ##2&-"S27 0'"'1 2'-,

Set heat block @65° C & Hyb oven @37° C Thaw Herring Sperm DNA & Yeast tRNA Add DIG Easy Hyb per tube 50 μl Mix by pipetting at least 50 x Centrifuge @ 13000 rpm 1 min Combine the 2 tubes (Alexa 555 & Alexa 647) Check dye incorporation on NanoDrop Add Herring Sperm DNA to prevent nonspecific binding 5 μl Add Yeast tRNA to prevent nonspecific binding 2 μl Incubate tube @ 65° C 5 min Centrifuge tube @ 13000 rpm 5 min Make slide bridge with 1× SSC in bottom of box Apply sample to 19K slide on slide bridge 95 μl Lower cleaned Lifter slip gently onto slide Cover dark slide box with foil Place in Hyb oven @37° C overnight 16 h ##2&-"(#62) 7S, 1

~~Prewarm second Hybridisation oven to 50° C Prepare the wash solutions as follows for one slide~~

~~200 ml 1× SSC (10 ml 20× SSC; 190 ml ddH20)~~

~~330~~

~~Appendices~~

~~150 ml 1× SSC; 0.1 % SDS (7.5 ml 20xSSC; 1.5 ml 10 %SDS; 141 ml ddH20)~~

~~50 ml 0.2× SSC (0.5ml 20× SSC; 49.5ml ddH20)~~

50 ml 0.1× SSC (0.25ml 20× SSC; 49.75ml ddH20) When the solutions and the oven have reached 50° C and the slide has been hybridising for 16 h, the washes can begin in the dark room 1× SSC @ RT to remove lifter slip (50 ml) 1 min 1× SSC 0.1%SDS @ 50° C oven on shaker (150 ml for 1 slide) 15 min 1× SSC @ 50° C oven on shaker (150 ml for 1 slide) 15 min 3rd wash of 1x SSC 0.1%SDS 20min The GenePix scanner is turned on to warm up the lasers during this 20 min wash cycle 0.2× SSC @ RT (50 ml for 1 slide) 1 min 0.1× SSC @ RT (50 ml for 1 slide) 1 min Slide is placed in a Falcon tube with kimwipe in the bottom and centrifuged @2000rpm for a few seconds The slide is now ready to be scanned

~~I.6 ANALYSIS OF MICROARRAY DATA USING BIOCONDUCTOR “R” STATISTICAL PACKAGE~~

Independent analysis of the raw data was carried out by Dr Rohan Williams (School of Biotechnology and Biomolecular Sciences, UNSW. Similar initial quality control measures such as exclusion of genes that had intensities less than the 95th percentile of the intensity values for the blank spots were used. Separate cutoff values were determined for each of the three sets of replicate arrays. Genes were also omitted from the analysis if complete data across all samples was not available. “R” is a freeware statistical computing environment (http://www.bioconductor.org) (Ihaka and Gentleman 1996) that normalises array data to remove biases when comparing intensity ratios from different arrays. The LIMMA package was used to perform print-tip LOWESS normalisation within arrays followed by further quantile-normalisation between the arrays. An “MA-plot” was used to visualise the (R = red for Cy5, G = green for Cy3) data. M is the log ratio of expression levels (log2(R / G)) and A is the log average of the expression levels (log2A(R×G)) for each gene. Statistical differential expression analysis (Dudoit, Yang et al. 2002) was computed using the linear models and empirical

~~331~~

~~Appendices~~

Bayesian methods in the LIMMA package (Smyth 2004). Bayes’s theorem is a way of quantifying uncertainty based on a probability theory using additional evidence and background information, leading to a number representing the degree of probability that the hypothesis is true. For the Bayesian model, it was assumed that approximately 1 % of expressed genes are differentially expressed. The B statistic was generated from t- tests to compare the low grade and high grade sarcoma lines and a moderated F statistic was generated for the three way comparison. Volcano plots of the two way comparison were created, plotting B values against the difference in M value between the high and low grade groups. Gene expression profiles for the top ranked genes from the comparison of the three groups were generated.

~~332~~

~~Appendices~~

~~"() 9~~

~~II. DIFFERENTIALLY EXPRESSED GENES, ONTOLOGIES and PATHWAYS on GENE EXPRESSION PROFILING of SARCOMA CELL LINES~~

~~333~~

~~Appendices~~

Table II.1 GeneSpring analysis comparing Low grade (SW684) and High grade (HT1080) cell lines Below is a selection of the 213 genes found to be over expressed in the high grade cell line by hierarchical clustering. The first 4 on the table were selected for real-time RT-PCR validation studies, as were the integrins ITGA2 and ITGA6. The list includes a number of genes involved in signal transduction, cell cycle regulation, cell adhesion and breakdown of extracellular matrix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

~~334~~

~~Appendices~~

Table II.2 R analysis comparing High and Low Grade Sarcoma Cell Lines List of 170 genes differentially expressed between high grade HT1080 and low grade SW684 with B statistic > 1 on R analysis (Bioconductor). The greater the B value, the better ranked the gene. The average difference in expression between the two cell lines (HT1080-SW684), is given as the change in M, which is a measure of the log2 ratio of red / green signal for each gene. A ΔΔΔM value of zero implies no difference in expression and ΔΔΔM = 1 implies a two fold difference. From this list, the expression of NDRG1, CCND1, ITGA2 and ITGA6 was validated using real time RT-PCR, as shown in Chapter 3.

($/$- 01 /2 2 " ($($ ( .(ΔΔΔ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

STZRY * V SZX[VR * V ,2'%#,.WW VUTYWSYW <RRSYZR * XU SWWYVR * XU ,2'%#,+#* ,-+ S ,2'%#, VSTZR[US <RSVUUW *(S XRWZ[V *(-. -.URR',&' '2-07.0-2#',S VSTXYXY 87RTVVYW 7"W.VUVTSX 7"W.VUVTSX.0-2#', VSTXYZVV

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

~~335~~

~~Appendices~~

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

~~<RRRSRY T XRRZSS " + %#V1.#!'$'! <8 ',"',%.0-2#',TQVZ) ST[X[RXU <RRXZYR 3< "#120', !2',"#.-*7+#0'8',%$ !2-0 VST[RWWS 8RSVWXX 88S XRXXTX "'1#**#" 11-!' 2#" !2'4 2-0 -$ TTRUVRST~~

~~336~~

~~Appendices~~

+-0.&-%#,#1'1S <RRVVS[ >.W XRURX[ "3 *1.#!'$'!'27.&-1.& 2 1#W TTUXYUYZ <RRS[ZW -3" SURVSR #*#!20-,V20 ,1$#0V$* 4-.0-2#',Q #2 VSTYSUY .-*7.#.2'"# <RRYRUX -S XRSWTS #,"-2&#*' *!#**V1.#!'$'!+-*#!3*#S TTZVRVXX <RSTURY -.VS/U XRWUUS #072&0-!72#+#+ 0 ,#.0-2#', ,"VTSV*')#U TTRZXUWX 8/U[RSVY "8TR* $ +'*75'2&1#/3#,!#1'+'* 0'27TRQ+#+ #0* VTTRRTWY 87RTX[RW *8/T $* 4-.0-2#',-6'"-0#"3!2 1#*8/T TTWSXTZY 8"TTWVTX "

~~337~~

~~Appendices~~

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

~~338~~

~~Appendices~~

! *'$-0,'! <RRRUYX : ( XRSYX[ 4'2 +', SQTWV "'&7"0-674'2 +', U VSTVUZX[ 0#!#.2-0 8?RR[U[[ F<3W XRXUXS 5',%*#11V27.# 3: ',2#%0 2'-, 1'2# $ +'*7Q VTTSSWYY +#+ #0W 8"RZWZW[ $3***#,%2&',1#02! <8!*-,#?ZVWTT 4#*-V! 0"'-V$ !' * 17,"0-+# TT/SS 0#%'-, TTTTZU[ +(<81#/3#,!# >[TZSX !UUTX 3,, +#" %-(:V%.0-2#',+(<8QST[XXYWZ !-+.*#2#!"1

~~339~~

~~Appendices~~

Table II.3 GeneSpring analysis of differentially expressed genes across the 3 tumour grades The symbols are those officially recognised by the Human Genome Organisation (HUGO) and the corresponding entry number for the gene in the Online Mendelian Inheritance in Man (OMIM) database, developed by the National Center for Biotechnology Information (NCBI) is also given. EGFR expression was validated using real time RT-PCR. ($/$- 01 /2 2 " ($($ (

~~<RRWWTW % SSS XRRYSU &7"0-6712#0-'"SSV #2 "#&7"0-%#, 1#S~~

~~<RSTTWW H(~~

~~<RRWTTZ -&"( SUSWWR #.'"#0+ *%0-52&$ !2-00#!#.2-0#072&0- * 12'!*#3)#+' 4'0 *4V#0 V -,!-%#,#&-+-*-%Q 4' , <RRTXUV .% SYXYRW.0-&''2',~~

~~<RRSYVY *8.& SWUXSW ! ..',%.0-2#', !2',$'* +#,2Q%#*1-*',V*')#~~

~~<RRU[WR "T(/U XRTYY[ !- %3* 2'-,$ !2-02&0-+ ',0#!#.2-0V*')#U~~

~~<RSWUSU 8(%&-"ST XRVYXU (&-%3 ,',#,3!*#-2'"##6!& ,%#$ !2-0&-"ST~~

~~STZRY * V SZX[VR * V ,2'%#,.WW~~

~~<RRR[RU <;6S STWZXR <8 .%"#&7"0-%#, 1#Q/3',-,#S~~

~~<RRSRR[ (.W XRUXUR 0' -1-+ *.0-2#',W~~

~~<RRYRRX *."W XRV[YZ !*# 4 %# ,".-*7 "#,7* 2'-,1.#!'$'!$ !2-0WQTW)~~

~~8DRRS[RV *(Z -5,17,"0-+#!0'2'! *0#%'-,%#,#Z~~

~~<RRTVTS .S STRUWU + 20'6+#2 **-.0-2#', 1#S',2#012'2' *!-** %#, 1#~~

~~USTST ?/X +7-1',Q*'%&2.-*7.#.2'"#XQ *) *'Q1+--2&+31!*# ,",-,V +31!*# <RRXZ[[ %U XRVWTX '1-!'20 2#"#&7"0-%#, 1#U<8 & #2~~

~~8/U[RSVZ *63 SSXY[R ! 2#!&-*V6V+#2&7*20 ,1$#0 1#~~

~~<RRRUWZ 3&" XRSX[T 20 ,1$-0+',%%0-52&$ !2-0Q #2 V',"3!#"QXZ)~~

~~<RRWV[Y &D8Y % .(3,!2'-,.0-2#',Q *.& YQVW) !-,,#6',VW~~

~~<RRRTRS *8S SVYZVR ',2#0!#**3* 0 "'-, +-*#!3*# S * WVQ &3+ , 0&',-4'0310#!#.2-0 <RRVZXV & "SW XRWUST %0-52&"'$$#0#,2' 2'-,$ !2-0SW~~

~~<RRTXWV .7T SY[RWR .7034 2#)', 1#Q+31!*#~~

SYSWT 6 T SVYVXR 13.#0-6'"#"'1+32 1#TQ+'2-!&-,"0' *

~~<RSZTUV 38.X "3"3*',T~~

~~<RSVVRT ;.V* *-5+-*#!3* 0+ 113 '/3',-,#V ',"',%.0-2#',[TW) ~~

~~340~~

~~Appendices~~

~~<RRSSTT 8 ". SRUS[W "'.-1#"'$$#0#,2' 2'-,V0#* 2#".0-2#',~~

~~<RRXRWZ 3<.S XRYYSV 3<"8.U',2#0 !2',%.0-2#',S~~

~~8"SXVY[Y (./SY +'2-!&-,"0' *0' -1-+ *.0-2#',/SY~~

~~<RSZYRR 3(UX 20'. 02'2#+-2'$V!-,2 ',',%UX~~

SUVWT /<8 SWRUUR* +',8 *

~~87RTUSW[ /SS >Y1,(<8V 11-!' 2#"+V*')#.0-2#',~~

~~<RRWWXZ /%HS XRS[[[/&-+#- -6S~~

~~8"RU[[VV < (&S XRWTXT~~

~~<RRWSXW 8/ 6* SRUZYR *"-* 1#*Q$03!2-1#V '1.&-1.& 2#~~

~~8/VVTRZX *TR-0$ZS !&0-+-1-+#TR-.#,0# "',%$0 +#ZS~~

~~8"RZXTUV *X-0$STW !&0-+-1-+#X-.#,0# "',%$0 +#STW~~

~~<RRSVSW -"TU URRSXS #3) 07-2'!20 ,1* 2'-,','2' 2'-,$ !2-0TQ13 3,'2U% ++ Q WT) <RSXVRU %.*SVZ &7.-2'! *.0-2#',%.*SVZ~~

~~87RTWVTU &*WTRSR &7.-2'! *.0-2#',&*WTRSR~~

~~<RRTURW /&8/S SWRWYR *#!2',Q% * !2-1'"#V ',"',%Q1-*3 *#QS% *#!2',S~~

~~<RRTUXV 8&-T URRR[Z +#* ,-+ ,2'%#,Q$ +'*7QT~~

~~<RRXWWX .:7 XRYXTT .&-1.&-+#4 *-, 2#)', 1#~~

~~87RTXWST 1'+'* 02- 8UYRXTS1'+'* 02-&'12-,#%T~~

~~WUXZYW 1'+'* 02-&'%&+- '*'27%0-3.83V&--)S~~

~~87RTSXTY W*F**T 8',!$',%#0Q*FV27.#5'2&!-'*#"V!-'*"-+ ',T~~

~~341~~

~~Appendices~~

Table II.4 Hierarchical Clustering Gene list of subtree shown in Figure 2.8 from hierarchical clustering indicating increasing levels of expression with increasing tumour grade. Functional groups of genes include those involved in embryonic development, signal transduction, cell cycle regulation and invasion. The gene symbols in the second column are those approved by the Human Genome Organisation (HUGO) and the third column provides the link to the Online Mendelian Inheritance in Man database (OMIM) developed by the National Center for Biotechnology Information (NCBI).

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

~~342~~

~~Appendices~~

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�', *.& V'+.-02', *.& U *')#*7 -02&-*-% -$ +-31# 1'%, **',% ',2#0+#"' 2# ', 3-** <RSXWZS 3.-* XRZUZZ . 2&5 7V#4-*32'-, 0'*7!-,1#04#" + ,,-17* *.& VSQUVV%*7!-.0-2#', #2 VSQVV3/S XRTYTV .# ,32V*')#S 0-1-.&'* <RRXUVY ..% XRXR[W .#.2'"7*.0-*7*'1-+#0 1#%!7!*-.&'*',% <RRVZSU .-HSX XRUUXR .#0-6'1-+ * '-%#,#1'1$ !2-0SX <RRWXZY "8(/ .&#,7* * ,',#V2(<817,2 1#V*')#Q #2 13 3,'2 <RSTTXZ ./ U .&-1.&-*'. 1# U <RRTXZY .<< XRUSWV .',',Q"#1+-1-+# 11-!' 2#".0-2#', 8"R[WTZZ .33&T XRVTUS .'23'2 0723+-0V20 ,1$-0+',%T

~~343~~

~~Appendices~~

<RRVZYW .6/(S* .-*7+#0 1#(<8.-*7.#.2'"#*QUR) <RSW[YT .6/(S .-*7+#0 1#(<8.-*7.#.2'"# QSX) >UYXZ[ .6/(T% XRXRTU .-*7+#0 1#(<8 <8"'0#!2#".-*7.#.2'"#% .-2 11'3+ ',5 0"*7V0#!2'$7',% !& ,,#*Q 13 $ +'*7 DQ <RRTTVU 7* .V .V <RSTTVU /*UW8U XRWXUT &*!<8!20 ,1.-02#0Q+#+ #08U <RRYUWW /*UWT 1-*32#! 00'#0$ +'*7UWQ+#+ #0T 1-*32# ! 00'#0 -0% ,'! ,'-, 20 ,1.-02#0 $ +'*7Q +#+ #0 <RSXUWV /*6V8S V8S <RRUSVX (.S XRVUTZ 1203!230#1.#!'$'!0#!-%,'2'-,.0-2#',S <RRRUVV &*.U 23 3*',Q% ++ !-+.*#6 11-!' 2#".0-2#',U <RRXT[T 3&SRS XRSUZY 23+-0131!#.2' '*'27%#,#SRS 8"RWXVUX !*-,#%-8SS*0'V"3V!& 2!0'2'! *0#%'-,+(<8 8"RZXVTS $3***#,%2&',1#02! <8!*-,#W YZ R[ ! <8 7"W.YXSSRSTS $0-+ !*-,# 8/SUYUSU 7"W.YXSSRSTS /UXWZY 1.*'!#">%&(<8 +(<8!-,2 ',',% 3,'/3#!#**3* 00#%'-,Q 7.,0#.# 2Q UTVVU ," >U&(0#%'-,-$3V*7+.&-20-.'!4'03127.#S

~~344~~

~~Appendices~~

Table II.5 Gene Ontologies for subtree obtained from hierarchical clustering This subtree showed increasing levels of gene expression with increasing grade (see also Figure 2.9 and Table 2.4 above for full gene list). The list of genes derived from the hierarchical clustering performed in GeneSpring (Silicon Genetics, Agilent Technologies, Palo Alto, CA) was uploaded to an online database, DAVID (http://apps1.niaid.nih.gov/david/), for analysis of the ontologies. )(2&1 $3 /(&-$($(0 "#,7*,3!*#-2'"# ',"',% ST 20 ,1!0'.2'-, SR !#**.0-*'$#0 2'-, [ + !0-+-*#!3*# '-17,2'1 [ 20 ,1.-02 [ .0-2#', '-17,2'1 Z !#**130$ !#0#!#.2-0*',)#"1'%, *20 ,1"3!2'-, Y ',20 !#**3* 01'%, **',%! 1! "# Y .0-2#',+-"'$'! 2'-, Y ("+#2 -*'1+ W .&-1.& 2#+#2 -*'1+ W .&-1.&-20 ,1$#0 1# !2'4'27Q *!-&-*%0-3. 1 !!#.2-0 W 0#1.-,1#2- '-2'!12'+3*31 W .0-2#',1#0',# 2&0#-,',#)', 1# !2'4'27 V %3 ,7*,3!*#-2'"# ',"',% U ,3!*#-2'"7*20 ,1$#0 1# !2'4'27 U .0-2#',2 0%#2',% U 0#1.-,1#2-.#12 . 2&-%#, . 0 1'2# U )("+#2 -*'1+ T V.0-2#',!-3.*#"0#!#.2-0 !2'4'27 T 1# !2'4'27 T +',- !'"+#2 -*'1+ T +',-13% 0+#2 -*'1+ T ! 0 -67*'! !'"+#2 -*'1+ T "#2#!2'-,-$#62#0, *12'+3*31 T %3 ,7*V,3!*#-2'"##6!& ,%#$ !2-0 !2'4'27 T &3+-0 *'++3,#0#1.-,1# T &7"0-* 1# !2'4'27Q !2',% -, !'" ,&7"0'"#1Q ', .&-1.&-031V T !-,2 ',',% ,&7"0'"#1 ',, 2#'++3,#0#1.-,1# T + !0-+-*#!3*#! 2 -*'1+ T ,#% 2'4#0#%3* 2'-,-$!#**.0-*'$#0 2'-, T ,#% 2'4#0#%3* 2'-,-$+#2 -*'1+ T ,3!*#-2'"#+#2 -*'1+ T -6'"-0#"3!2 1# !2'4'27Q !2',% -, 2&# 2V 2 %0-3. -$ "-,-01Q T (")-0(") 1 !!#.2-0 .0-%0 ++#"!#**"# 2& T .0-2#',! 2 -*'1+ T .0-2#',$-*"',% T .0-2#',V270-1',#)', 1# !2'4'27 T 0#%3* 2'-,-$ '-17,2'1 T 0#1.-,1#2-)("" + %#12'+3*31 T 0#1.-,1#2-5-3,"',% T 1#,1-07.#0!#.2'-, T 20 ,1!0'.2'-,$ !2-0 !2'4'27 T 20 ,1$#0 1# !2'4'27Q20 ,1$#00',%-17*%0-3.1 T 8',!'-, ',"',% T 3,!* 11'$'#" WW

~~345~~

~~Appendices~~

Table II.6 Set 1 from k means clustering These genes are more highly expressed in the high grade and metastatic cell lines than the low grade. There is good correlation between this list of genes and that obtained through hierarchical clustering, using the portion of the gene tree showing higher expression in the high grade and metastatic lines, as shown in Table 2.6. Each “set” in k means represents a cluster of genes with common expression patterns. The gene symbol approved by the Human Genome Organisation (HUGO) Gene Nomenclature Committee appears in the second column. The citation number for the corresponding entry for the gene in the Online Mendelian Inheritance in Man (OMIM) database, developed by the National Centre for Biotechnology Information (NCBI), where present, is given in the third column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

~~346~~

~~Appendices~~

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

~~347~~

~~Appendices~~

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

~~348~~

~~Appendices~~

~~8DRSSV[Y */~~

~~349~~

~~Appendices~~

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�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

~~350~~

~~Appendices~~

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

~~351~~

~~Appendices~~

>UYXZ[ .6/(T% XRXRTU .-*7+#0 1#(<8 <8"'0#!2#".-*7.#.2'"#% <RRXVXY (.*UT .-*7+#0 1#(<8 <8"'0#!2#"UT) <RRWRUW .6/(3 XRSYYZ .-*7+#0 1#(<8+'2-!&-,"0' * <8"'0#!2#" <RRTTVU 7*

~~352~~

~~Appendices~~

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

~~353~~

~~Appendices~~

<RRSTWR 3<"("W SR[WUW 23+-0 ,#!0-1'1 $ !2-0 0#!#.2-0 13.#0$ +'*7Q +#+ #0W <RRWRY[ 3. WT XRVRXZ 23+-0.0-2#', WT 87RTXSV[ 3>*U XRSUZW 23+-013..0#11-0! ,"'" 2#U <RRXT[T 3&SRS XRSUZY 23+-0131!#.2' '*'27%#,#SRS <RRUUUU >8WT S[SUTS 3 '/3'2', 8VWT 0#1'"3# 0' -1-+ * .0-2#', $31'-, .0-"3!2S 8/SSRSZW >&W 3.0#%3* 2#""30',%1)#*#2 *+31!*#%0-52&W <RRU[RV W<"TW[ XRU[RS 8',!$',%#0.0-2#',TW[ 8RVR[SZ W<"VRX 8',!$',%#0.0-2#',VRX

~~354~~

~~Appendices~~

Figure II.1 Gene Ontology from k means Cluster Set 1 The gene ontologies for Set 1 shown in the previous image of k means clustering were analysed and are represented on the pie chart below. The greatest proportion of genes fall into the categories of physiologic process, binding and cellular process when the least stringent criteria are used for the categorisation. This provides the greatest coverage, that is, the greatest number of genes are classified but the categories used are least specific. More stringent criteria were used for the ontology shown in Figure 2.7, with a far greater number of defined categories. However, general patterns can be discerned from the chart below, with a significant proportion of genes being involved in signal transduction and transcriptional regulation. Genes that should only be expressed during development or embryogenesis are also represented, indicating an aberrant activation of these pathways during tumorigenesis.

~~GO annotations for k means clustering Set 1~~

physiological process binding cellular process catalytic activity signal transducer activity structural molecule activity transporter activity development regulation of biological process transcription regulator activity molecular_function unknown biological_process unknown translation regulator activity chaperone activity enzyme regulator activity unclassified

~~355~~

~~Appendices~~

Table II.7 Set 3 from K means clustering Genes in Set 3 with highest levels of expression in the metastatic group (GCT), as shown in Figure 2.11. The gene symbols in the second column are those approved by the Human Genome Organisation (HUGO) Gene Nomenclature Committee and the third column provides the corresponding citation number for the gene in the Online Mendelian Inheritance in Man (OMIM) database, developed by the National Centre for Biotechnology Information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

~~356~~

~~Appendices~~

~~<RSYWVR &8/<3SR XRZRVU > .V~~

~~357~~

~~Appendices~~

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

~~358~~

~~Appendices~~

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

~~359~~

~~Appendices~~

8"RUZ[WS W<"TWU XRX[WV 8',!$',%#0.0-2#',TWU 8"RRSZ[T 8"RTVYRZ 8"RWXVTT 8"RYWRZX 8"RZXTRV 8DTTYZZ[ 8DTTYZ[T 87RRTSXT 87RTTUW[ 8/RV[TWT 8/SUYVTZ 8/UW[XRY 8?RRYSTZ SX[SV YRZUV /SRSVR ZY[VU <RRSYTU <RRS[W[ <RRUZYS <RSXTWU WUXY[X

~~360~~

~~Appendices~~

Table II.8 R analysis of genes differentially expressed across the three tumour grades The moderated F statistic is used as method of gene ranking. The top 60 differentially expressed genes are presented here. As for the previous tables, annotations were performed by uploading the gene list to an online database, DAVID (http://apps1.niaid.nih.gov/david/), accessed through the National Institute of Allergy and Infectious Diseases (NIAID) website. This annotation tool was used to provide the gene name, together with the gene symbol approved by the Human Genome Organisation (HUGO) Gene Nomenclature Committee, the link to the entry in the Online Mendelian Inheritance in Man (OMIM) database, as well as to Entrez Gene (not shown), these last two, developed by the National Centre for Biotechnology Information (NCBI). ($/$- 0 01 /2 2 " ($($ ( <RRSS[X YTXYTXVY XRS[[Y %U',2#0 !2',%"-+ ',"# 2& %-,'12 STZRY YTUTTXZ[ * V SZX[VR * V ,2'%#,.WW 8DRRS[RV XTTXVZS *(Z -5,17,"0-+#!0'2'! *0#%'-,%#,#Z <RRVSRW ZTWRYXRU -"-.S XRSWVZ -&"V!-,2 ',',% $' 3*',V*')# #620 !#**3* 0 + 20'6.0-2#',S <RTRVSS STTWWUZT &8&- T URRTZ[ & ,2'%#,Q$ +'*7 QT <RRSVZY XTYUUWZX &*

~~361~~

~~Appendices~~

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

~~362~~

~~Appendices~~

~~a b~~

Figure II.2 The Notch pathway This ancient signalling system regulates cell fate and initiates differentiation in normal embryogenesis. Aberrant Notch signalling has been implicated in both tumour suppression and tumour progression. Insert (a) shows a normally developed Drosophila melanogaster wing, while insert (b) demonstrates the “notch” that results from deficient function of the Notch gene. Notch4 was overexpressed in the high grade and metastatic cell line in this study, as was a related receptor, DNER (Delta- and Notch-like EGF related receptor). Both the ligands and the Notch receptors contain a variable number of epidermal growth factor-like repeates. DNER comprises an extracellualr domain with 10 EGF-like repeats and an intracellular domain with tyrosine kinase activity. Notch signalling also modulates the MAPK pathway. This figure was adapted from the Kyoto Encyclopaedia of Genes and Genomes or KEGG database (http://www.genome.jp/kegg/pathway.html).

~~363~~

~~Appendices~~

Figure II.3 The Integrin Pathway Extracellular matrix (ECM) components such as laminin bind to the transmembrane heterodimeric integrins (ITGA-ITGB, circled red). Focal adhesion kinase (FAK) is activated and multiple signalling cascades are triggered. This eventually leads to promotion of cell survival and proliferation, as well as changes necessary for cell migration (purple boxes). This gene expression study showed progressive increase in expression levels of Integrin alpha 2 and 6 (ITGA2, ITGA6) and Laminin B3 across the cell lines of increasing tumour grade. Certain matrix metalloproteinases (MMPs), which may be regulated by transforming growth factor beta (TGFB) via Smad- independent pathways, as well as by integrins, were also overexpressed in the high grade and metastatic sarcoma lines. This figure was adapted from the Kyoto Encyclopaedia of Genes and Genomes or KEGG database (http://www.genome.jp/kegg/pathway.html).

~~364~~

~~Appendices~~

~~Smad Independent Smad Independent Mechanisms~~

~~Angiogenesis Cell Motility Endoglin & Invasion ALK1 Integrins Cell Resistance to Proliferation TGFB Ras/Raf/ mediated MAPK A t i (?)~~

~~Immune Genome Suppression Instability ↓IL2, T-cell Prevents DNA ti ti repair~~

Figure II.4 Transforming growth factor beta (TGFB) functions TGFB has both tumor-suppressor and tumor-promoter functions (a, left panel). One theory holds that the positive effects on tumour progression occur through stromal cells, which are of course, mesechymal derived. The other model holds that the phenotypic switch that occurs in the epithelial tumour (EMT), changes its response to TGFB. These tumour promoter functions may be effected through Smad-independent mechanisms (a, right panel and b), including induction of angiogenesis via the TGFB receptor on endothelial cells. In the current gene expression profiling study, TGFB1, ACVR2B, p38 MAPK and TAK1 were all overexpressed, suggesting that in STS, which are mesenchymal tumours, TGFB may similarly function through Smad independent means, effecting the same results. The left panel of figure (a) was reproduced from Elliot and Blobe, 2005 and (b) from Derynck, R. and Zhang, Y. E., 2003.

~~365~~

~~Appendices~~

~~"() 9~~

~~III. ADDITIONAL PROTOCOLS and STATISTICAL ANALYSIS of VALIDATION of GENE EXPRESSION DATA using qRTPCR~~

~~366~~

~~Appendices~~

~~III.1 Rationale for validation studies in Gene Profiling~~

Several controversies remain as to the sensitivity and specificity of gene expression arrays, such as annotation errors, experimental variability and cross hybridisation of “non target” transcripts (King and Sinha 2001). The validation process can be divided into a number of components: quality control of experimental variability, independent verification of results on the same samples and assessing the universality of the results (Chuaqui, Bonner et al. 2002). The first of these was implemented in Chapter 2, dealing with standardisation of microarray experimental protocols, the use of replicates and quality control measures in the data analysis. Chapter 3 deals with the second component: verifying the results on the same samples, using quantitative RT-PCR in real time (qRTPCR). The third component, universality, is considered in Chapter 4 on tissue microarrays (TMA), where immunohistochemistry (IHC) is used for the evaluation of protein expression levels on a large number of patient samples.

Previous gene expression studies on soft tissue sarcomas have used a variety of methods for confirmation of gene expression microarray data, for instance, Khan et al in 1999 (Khan, Bittner et al. 1999) verified a number of genes using Northern analysis, while Ren et al 2003 (Ren, Yu et al. 2003) used traditional RT-PCR with ACTB (β-actin) as the internal control and Nagayama et al 2002 (Nagayama, Katagiri et al. 2002) validated their results using real-time RT-PCR (qRTPCR) with β2 microglobulin as the housekeeping gene.

The potential benefits of applying the technique of real time RT-PCR in clinical practice is already being appreciated. The feasibility of using GD2 expression in neuroblastoma as a molecular marker for the disease was assessed (Cheung and Cheung 2001). GD2 expression correlated well with prognosis, illustrating the diagnostic and prognostic potential of the assay in clinical practice. Protocols for obtaining intact mRNA from renal biopsies for evaluation of transcripts using qRTPCR in the clinical setting have been established (Cohen, Frach et al. 2002). More recent studies have evaluated mRNA expression of biomarkers on a large number of soft tissue sarcoma specimens, performing clinical correlations at the same time (Oda, Saito et al. 2005; Taubert, Kappler et al. 2005).

~~367~~

~~Appendices~~

~~III.2 Advantages of Real Time PCR~~

Fluorescence, which is proportionate to the amount of amplified product, is measured during the exponential phase of amplification, allowing for greater dynamic range of measurement and collection of data (Heid, Stevens et al. 1996). During this log phase, product the reaction is most efficient, with least variability. The threshold cycle (Ct) is the cycle number at which the exponential increase in signal is detected. The higher the mRNA copy number of the target, the sooner it enters the exponential phase of amplification, and consequently, the earlier the fluorescence signal is detected (Bustin 2000). Visualisation of the reactions during amplification also acts as a quality control measure. In a study that compared the techniques of primer dropping43, competitive44 and qRTPCR (Wall and Edwards 2002), qRTPCR compared favorably with nested competitive RT-PCR in terms of sensitivity, linearity, and reproducibility. Another advantage cited by the authors was the greater detection range for target cDNA.

~~III.3 Normalisation~~

43 Primer dropping refers to a technique of normalisation where the normalising gene is amplified within the same reaction as the target gene, but the primers for the normalising gene are added to the reaction after the first few cycles of the PCR. 44 Competitive RT-PCR uses a synthetic RNA within the same reaction as the sample RNA

~~368~~

~~Appendices~~

correction of variables such as different amounts or quality of starting mRNA, as well as differences in cDNA synthesis, with the same starting RNA and cDNA template being used to assay both the target genes and the HKG. For the purpose of this or indeed any study therefore, the choice of internal control or HKG should be one that has constant levels of expression for the samples being tested.

HKG or non-regulated genes such as GAPDH, ACTB (β-actin), 18S and B2M (β2 Microglobulin) have traditionally been used in real-time PCR assays. There is, however, debate over the choice of HKG on the premise that these so called HKG do actually have variable expression levels, depending on the tissues studied (Bustin 2000; Suzuki, Higgins et al. 2000; Vandesompele, De Preter et al. 2002; Radonic, Thulke et al. 2004). Moreover, the expression may vary according to experimental procedure. A previous report indicated stable expression of GAPDH and 18S rRNA in various types of neural tissue but variable GAPDH expression in cultured peripheral blood mononuclear cells when stimulated by mitogen activators added to the culture media (Thellin, Zorzi et al. 1999).

~~Therefore in the present study, two HKGs, GAPDH and ACTB, both known to be ubiquitously and abundantly expressed in most tissues were assayed on each experiment.~~

~~III.4 Relative Quantitation~~

There are two methods of relative quantitation that can be applied in real time RT-PCR. The first is the standard curve method, which requires serial dilutions of cDNA template for each target transcript to be amplified with each experiment to generate the standard curve.

~~III.4.1 The Comparative Ct Formula~~

The second, is the comparative Ct method (Pfaffl 2001), which as indicated in the previous section, normalises the expression of the target transcript against a housekeeping gene and compares the expression of the transcript in the sample to a control. This latter method allows for greater experimental throughput as multiple wells are not required to generate a standard curve. For the mathematical model used in this

~~369~~

~~Appendices~~

method to be applied, the amplification efficiencies of the target and normaliser have to first be determined. The mathematical calculations involved and the derivation of a numerical value for the efficiency of the amplification are described in greater detail in Chapter 3, Section 3.2.6. The validity of using the comparative Ct formula for relative quantitation was confirmed by assessing the amplification efficiencies of target housekeeping transcripts. As the efficiencies for both HKG and candidate genes were close to 2, the formula could be applied. However, as this required doing separate calculations to normalize against each reference gene, the REST© 2005 program was used.

The relative quantitation of gene expression can also be analysed with the same control sample and housekeeping genes using the Relative Expression Software Tool (REST 2005 v1.9.10) (Pfaffl, Horgan et al. 2002). The mathematical model used in this program takes into account the efficiency of the PCR reaction for each candidate gene. In addition, normalisation to more than one reference gene can be carried out by calculating a geometric mean for the reference genes, using this to compute a normalisation factor (Vandesompele et al and Pfaffl 2002). The use of the statistical model, known as the Pair Wise Fixed Reallocation Randomisation Test with 50000 iterations, is useful as it does not assume normal distribution of data points as do other standard parametric tests such as t-tests. For these reasons, the raw data was re-analysed using this program, providing expression ratios for each sarcoma cell line (to the control cell line MRC5). The ratios were then examined for statistical significance.

~~III.5 Housekeeping/Reference Genes~~

GAPDH and ACTB were used as the housekeeping genes (HKG) based on a review of the literature on soft tissue sarcoma where real time RT-PCR was utilised as a technique for evaluation of mRNA transcripts. GAPDH was the reference gene when survivin expression was evaluated in 56 soft tissue sarcomas (STS) (Kappler, Köhler et al. 2001). HER-2/neu (ERBB2) expression in synovial sarcoma (SS), osteosarcoma and Ewing’s sarcoma (Thomas, Giordano et al. 2002; Thomas, Giordano et al. 2005) and ERBB3 expression in clear cell sarcoma cell lines were examined (Schaefer, Brachwitz

~~370~~

~~Appendices~~

et al. 2004). In all three studies GAPDH was the HKG. The expression of Snail in SS was also examined using GAPDH as the HKG (Saito, Oda et al. 2004). In a study involving Ewing’s sarcoma, synovial sarcoma and alveolar rhabdomyosarcoma, among others, two housekeeping genes GAPDH and β2 microglobulin were employed (Hill, O'Sullivan et al. 2002). It is however known that GAPDH mRNA levels can fluctuate, given that it is involved in a variety of cellular functions such as DNA repair and replication, microtubule bundling and phosphotransferase activity (Sirover 1999). Nonetheless, in an experiment comparing methods of qRTPCR for expression of matrix metalloproteinase 13 (MMP13), there was no difference between normalisation with GAPDH or 18S (Wall and Edwards 2002). GAPDH expression increased, however, in vitro in hypoxic conditions, oxidative stress or with exposure to transition metals (Suzuki, Higgins et al. 2000). Cellular proliferation elevated GAPDH expression, suggesting cell cycle dependant regulation of GAPDH. It follows that GAPDH expression has been shown to be constitutively increased in some transformed cells, due to the increased glycolytic activity and proliferation rate in malignant cells. This may explain the increased level of expression of GAPDH seen in the three sarcoma cell lines (SW684, HT1080 and GCT) compared to the normal lung fibroblast cell line MRC5 in this study. In contrast to GAPDH, there were far fewer instances where ACTB expression was modulated (Suzuki, Higgins et al. 2000). Numerous studies involving sarcoma utilised ACTB as the reference gene (Noguchi, Ueki et al. 1997; Frolov, Chahwan et al. 2003; Nuciforo, Pellegrini et al. 2003; Ren, Yu et al. 2003), although for some of these studies RT-PCR was not carried out in real time and hence, relative quantitation was not performed (Frolov, Chahwan et al. 2003; Ren, Yu et al. 2003). In the present study, ACTB expression was lower in the sarcoma cell lines compared to the normal fibroblast cell line. However, it should be noted that MRC5 was the control cell line used in both the gene expression arrays in Chapter 2 (common reference design) and the qRTPCR experiments in this chapter and that the comparison being carried out is actually that across the three sarcoma cell lines. On one way ANOVA, the expression of the reference genes across the sarcoma cell lines was not significantly different.

~~371~~

~~Appendices~~

The drawback of using a single reference gene was considered at the outset of this study but using a large number of reference genes would be prohibitive in terms of resources. Thus both GAPDH and ACTB were used together as reference genes in the Relative Expression Software Tool (REST© 2005), calculating the geometric mean of these.

~~372~~

~~Appendices~~

~~III.6 ADDITIONAL RESULTS~~

~~III.6.1 PCR Efficiency~~

~~&8. % '*32'-, #0'#1 8+.*'$'! 2'-,*304#1~~

<3*

~~&8. % 8+.*'!-,1~~

Figure III.1 PCR Dilution Series and Melt Curves for GAPDH The amplification curves for the dilution series of pooled cDNA (Neat, 1:10, 1:50, 1:100, 1:500, 1:1000, each in duplicate) are shown in graph (a), together with that of the non template control (NTC). Product amplification commences at progressively later cycle numbers (x-axis) the more dilute the template. Replicate samples group together. The efficiency of the reaction is calculated, by plotting a graph of the threshold cycle, Ct, against the Log of the concentration of cDNA. The corresponding melt curves are shown in (b). Melting occurs at the same temperature but the peaks are lower, indicating less total product. However, as GAPDH is abundantly expressed, this product gradient is less obvious than in the melt curves shown for EGFR.

~~373~~

~~Appendices~~

~~NTC~~

~~GAPDH~~

~~CCND1~~

~~NTC~~

Figure III.2 PCR Dilution Series and Melt Curves for CCND1 and GAPDH The amplification curves for the dilution series of pooled cDNA (Neat, 1:2, 1:10, 1:50, 1:100, each in duplicate) as carried out on the Rotor-Gene® 3000 Sequence Detection System (Corbett Life Science, Sydney, Australia) are shown in graph (a), with product amplification commencing at progressively later cycle numbers (x-axis) the more dilute the template. This dilution series is used to calculate the efficiency of the reaction, by plotting a graph of the threshold cycle, Ct, against the Log of the concentration of cDNA, as shown in Figure III.3, Appendix III. The corresponding melt curves are shown in (b). Melting occurs at the same temperature but the peaks are lower, indicating less total product. NTC refers to the non template control.

~~374~~

~~Appendices~~

~~PCR Ef f i ci e ncy: ACV R2B~~

~~t 20~~

~~Mean C 15 y = -3.1323x + 30.98 R2 = 0.9942 10~~

~~0 -0.5 0 0.5 1 1.5 2 2.5 3 3.5 Log (cDNA Conc. ng/ul)~~

~~ACVR2B Linear (ACVR2B)~~

~~10 y = -3.168x + 21.903 R2 = 0.9948 5~~

~~Mean Ct (GA PDH) Linear (Mean Ct (GAPDH))~~

Figure III.3 PCR Efficiency plots of (a) ACVR2B and (b) GAPDH This plot of the Mean Ct values against the Log (cDNA template concentration) is used to determine whether the relative quantitation method as described by Pfaffl can be utilised for the given target genes and chosen housekeeping gene (GAPDH). The slope of each line is used to calculate the efficiency (E) of the amplification, using the formula E = 10^(-1/Slope). If the efficiency E -(ΔΔΔΔΔΔCt) target gene = E housekeeping gene = 2, then using the formula as given by Michael Pfaffl’s, 2^ , is justified. From the above examples, where a dilution series of pooled cDNA from all the cell lines was used, EACVR2B = 2.09 and EGAPDH = 2.07. Similar values were obtained for other target genes. The Pfaffl formula was therefore used for subsequent relative quantitation.

~~375~~

~~Appendices~~

~~y = -4.4909x + 28.733 R2 = 0.9835~~

~~Mean Ct~~

~~Series1 Linear (Series1)~~

~~y = -3.8394x + 25.841 R2 = 0.9971~~

Series1 Linear (Series1) Figure III.4 PCR Efficiency plots of (a) ITGA2 and (b) ITGA6 This plot of the Mean Ct values against the Log (cDNA template concentration) is used to determine whether the relative quantitation method as described by Pfaffl can be utilised for the given target genes and chosen housekeeping gene (GAPDH). The slope of each line, as calculated from the line of best fit, is used to calculate the efficiency (E) of the amplification, using the formula E = 10^(-1/Slope). From the above examples, where a dilution series of pooled cDNA from all the cell lines was used, EITGA2 = 1.89 and EITGA6 = 1.82. Similar values were obtained for other target genes run on Rotor-Gene® 3000 Sequence Detection System (Corbett Life Science, Sydney, Australia), including the reference genes.

~~376~~

~~Appendices~~

~~y = -4.6345x + 27.841 R2 = 0.9595~~

~~Series1 Linear (Series1)~~

~~y = -3.8447x + 25.168 R2 = 0.9753~~

CCND1 Linear (CCND1) Figure III.5 PCR Efficiency plots of (a) NDRG1 and (b) CCND1 The slope of each line, as calculated from the line of best fit, is used to calculate the efficiency (E) of the amplification, using the formula E = 10^(-1/Slope). From the above examples, where a dilution series of pooled cDNA from all the cell lines was used, ENDRG1 = 1.84 and ECCND1 = 1.82. Similar values were obtained for other target genes run on Rotor-Gene® 3000 Sequence Detection System (Corbett Life Science, Sydney, Australia), including the reference genes.

~~377~~

~~Appendices~~

~~III.6.2 Differential Transcript Expression~~

Table III.1 Differential Expression Between Cell Lines 0 ,1!0'.2 F ## ,)'$$#0#,!# VF .4 *3# 3&8X FXZV &2SRZR VUTYUWXXW# RTRRS %!2 VSTRXZWST# RTRVV %3SRZR 15XZV UTYUWXXW# RTRRS %!2 TTXXYSWU# RTRRU &*3 15XZV STRXZWST# RTRVV &2SRZR VTTXXYSWU# RTRRU < (&S FXZV &2SRZR VVTXYWRUR# RTRRY %!2 VSTU[RSXU RTS[T %3SRZR 15XZV VTXYWRUR# RTRRY %!2 UTTZVZXX# RTRTR &*3 15XZV STU[RSXU RTS[T &2SRZR VUTTZVZXX# RTRTR 3&8T FXZV &2SRZR VUTTRS[ZY# RTRRW %!2 VRT[WYWVY RTV[S %3SRZR 15XZV UTTRS[ZY# RTRRW %!2 TTTVVVVR# RTRUR &*3 15XZV RT[WYWVY RTV[S &2SRZR VTTTVVVVR# RTRUR -&"( FXZV &2SRZR VTTT[UYRY# RTRRT %!2 VTTYX[WS[# RTRRS %3SRZR 15XZV TTT[UYRY# RTRRT %!2 VRTVYWZST RTYU[ &*3 15XZV TTYX[WS[# RTRRS &2SRZR RTVYWZST RTYU[ **< S FXZV &2SRZR VSTUYUSWW# RTRRS %!2 VSTTSYXWV# RTRRU %3SRZR 15XZV STUYUSWW# RTRRS %!2 RTSWWWRS STRRR &*3 15XZV STTSYXWV# RTRRU &2SRZR VRTSWWWRS STRRR 8*:(T FXZV &2SRZR VRTVSS[UV RTW[[ %!2 VT[YWSWW# RTRT[ %3SRZR 15XZV RTVSS[UV RTW[[ %!2 VRTWXUTTS RTTYU &*3 15XZV T[YWSWW# RTRT[ &2SRZR RTWXUTTS RTTYU 3&#"'$$#0#,2' *#6.0#11'-, #25##,# !&!#***',#$-02&-1#! ,"'" 2#%#,#1$-3,"2- #1'%,'$'! ,2 -,3 *#UTT'1.0#1#,2#" -4#T'%,'$'! ,2"'$$#0#,!#1 0# 12#0'1)#"#T

~~378~~

~~Appendices~~

~~"() 9 +~~

~~IV. TISSUE MICROARRAY: VALIDITY, PROTOCOLS and DETAILED DATA TABLES~~

~~379~~

~~Appendices~~

~~IV.1 Validity of Tissue Microarray (TMA)~~

Two concerns raised with regard to the validity of the technique was firstly, that the 0.6 mm cores taken may not be representative of the entire tumour and secondly, that archival specimens that may have lost their antigenicity would be used.

~~IV.1.1 Tumour heterogeneity~~

Traditional standards of histologic examination dictate that tissue should be sampled at the rate of one section per cm3 of tumour. Each 0.6 mm core represents only about 1.4 x 106 μm3, about 0.3% of the tissue traditionally considered by pathologists to be “representative” (Camp, Charette et al. 2000). A number of validation studies have addressed these issues (Table 4.1). Many of these were carried out on breast cancer, using previously validated markers such as oestrogen and progesterone receptor (ER and PR) status, HER2 and p53 (Camp, Charette et al. 2000; Gillett, Springall et al. 2000; Torhorst, Bucher et al. 2001; Skacel, Skilton et al. 2002). The aim was to determine the optimal number of cores. Both the leading edge and the centre of the tumours were sampled, based on the premise that expression patterns of markers may differ at these locations (Torhorst, Bucher et al. 2001). This premise was also examined in colorectal cancers (CRC) (Jourdan, Sebbagh et al. 2003). Neither study found any difference between cores sampled from the leading edge or from the centre of the tumour. Analysis of a single readable core matched the staining pattern of an entire section more than 90% of the time. The highest rates of concordance for 3 cores per tumour were 98.8-99.6% (Jourdan, Sebbagh et al. 2003).

Ki-67 expression in two prostate cancer TMAs was investigated (Rubin, Dunn et al. 2002). The optimal number of cores was determined to be three. A greater number did not improve the accuracy of the results. The optimal size of cores was also examined, hypothesising that larger cores may be more representative (Skacel, Skilton et al. 2002). However, almost equivalent results were obtained for both 1.5 mm and 0.6 mm cores. Mucci et al employed TMA as a screening tool to evaluate the prevalence of neuroendocrine (NE) expression in prostate cancers, which had been estimated to range from 5 - 83% (Mucci, Akdas et al. 2000). It was hypothesised that sampling small cores

~~380~~

~~Appendices~~

of tissue on a tissue array would not accurately represent the tumour given the focal expression of the markers investigated (chromogranin A and synaptophysin). It was found that TMA, with its capacity for examining a large number of samples under the same staining conditions, could better estimate the probability of focal expression of NE markers. The concept of using TMAs for “tumour screening” had also been investigated in 17 types of epithelial tumours evaluated for amplifications of CMYC, CCND1 and ERBB2 (Schraml, Kononen et al. 1999). The TMA findings agreed with the published literature in 73% of cases.

~~IV.1.2 Antigenicity of archival specimens~~

Tissues embedded in paraffin are thought to be protected from oxidative or other damage. However on sectioning paraffin blocks, tissues may lose their antigenicity (Jacobs, Prioleau et al. 1996; Shin, Kalapurakal et al. 1997). Some of the TMA validation studies have sought to address this concern, with specimens sourced from the 1930s to present (Camp, Charette et al. 2000; Dolled-Filhart, Camp et al. 2003) (Table 4.1). There was no significant difference in staining across the decades, with the exception of oestrogen receptor (ER) staining for specimens from the 1930s. Either the tissues had lost their antigenicity, or the prevalence of ER-positive tumours was significantly different in the 1930s. Dolled-Filhart et al hypothesised that archival tissue may lose antigenicity for phosphorylated antibodies, and tested this by performing immunostaining on their TMAs for phosphorylated and unphosphorylated STAT3. Positive staining was achieved for both STAT3 and pSTAT3.

~~IV.1.3 Inter- and intra-laboratory variability and reproducibility~~

Immunostaining is subject to local variations in tissue fixation, antigen retrieval and staining. TMAs may in fact reduce variability by allowing standardisation of experiment conditions on all samples. Inter laboratory and interobserver reproducibility was examined in a large multi-centre trial involving 172 laboratories, all of whom received sectioned TMA slides (von Wasielewski, Mengel et al. 2002). The greatest concordance was found for tissue cores staining strongly (82-97%).

~~381~~

~~Appendices~~

~~IV.1.4 Validation in soft tissue tumours~~

Some investigators have validated the use of this technology in soft tissue tumours. An early study examined 11 STS, all of the malignant fibrous histiocytoma (MFH) subtype, on the grounds that MFH represents a heterogeneous group (Engellau, Akerman et al. 2001). Good correlation was achieved between TMAs and conventional sections. Other studies have examined fibrosarcomas and desmoid tumours (Hoos, Urist et al. 2001) for three markers (Ki-67, p43, Rb). The rates of tissue loss from the arrays due to the sectioning and staining processes ranged from 10% to 20%. 91-98% concordance was attained when comparing immunostaining of the arrays to whole tissue sections. Interlaboratory reproducibility was investigated using a multitumour array and a panel of 22 antibodies (Hsu, Nielsen et al. 2002). The immunostaining profile for their tumours was largely in agreement with the published literature, where available.

~~IV.2 Application of tissue microarray technology~~

While TMA technology lends itself to any entity where paraffin-embedded archival tissue is available, the earliest and most numerous application of it was in the study of cancer. Correlation of immunostaining patterns with data such as tumour grade or survival allowed for statistical evaluation of independent prognostic variables.

~~IV.2.1 TMA in epithelial malignancy~~

Table 4.2 summarises selected studies carried out on epithelial tumours using TMA. Some have sought to validate biomarkers or gene amplifications found to be significant on cDNA- or comparative genomic hybridisation (CGH) arrays (Barlund, Forozan et al. 2000; Hedenfalk, Duggan et al. 2001; Monni, Barlund et al. 2001). Biomarkers identified in this fashion such as IGFBP2, and S100P were also examined on prostate cancer specimens on TMAs (Bubendorf, Kononen et al. 1999; Mousses, Bubendorf et al. 2002). The expression of hepsin and pim-1 in a prostate cancer TMA was correlated with clinical outcome (Dhanasekaran, Barrette et al. 2001).

Vimentin expression on a tissue microarray of renal cell carcinomas of varying grades was used to correlate vimentin expression with tumour grade and prognosis (Moch, Schraml et al. 1999). A significant difference in IGFBP2 expression between low and high grade astrocytomas has also been demonstrated (Sallinen, Sallinen et al. 2000).

~~382~~

~~Appendices~~

Bladder cancers have also been evaluated using tissue array technology (Richter, Wagner et al. 2000; Sanchez-Carbayo, Socci et al. 2002). Expression of E-cadherin, zysin and moesin correlated with tumour grade and moesin expression correlated with outcome (Sanchez-Carbayo, Socci et al. 2002).

The gene expression pattern of a non small cell lung cancer (NSCLC) cell line was examined using CGH and spectral karyotyping (SKY) (Tai, Sham et al. 2006). Amplification of EGFR and FGF3 were identified and then examined at the level of protein expression on a TMA of NSCLC samples. However, there was no correlation between tumour type or stage and no survival analysis appeared to have been carried out.

TMAs have also been utilised to examine the relationship between tumour phenotype and mechanism of action. Potential activation of the JAK-STAT signal transduction pathway has recently been evaluated in a TMA comprising matched samples of primary and recurrent or recurrent metastatic ovarian cancer (Duan, Foster et al. 2006), finding a trend toward increased expression of phosphorylated STAT3 (pSTAT3) in the recurrent malignancy specimens.

~~IV.2.2 TMA in soft tissue sarcoma~~

TMA studies involving STS are summarised in Table 4.3. The early studies found characteristic expression profiles for the sarcomas of simple karyotype such as synovial sarcoma (SS) (Allander, Illei et al. 2002; Nagayama, Katagiri et al. 2002; Segal, Pavlidis et al. 2003). Allander et al described a characteristic gene expression profile for SS and then examined two overexpressed factors on clinical samples on a TMA (Allander, Illei et al. 2002). Other TMA studies on SS and gastrointestinal stromal tumours (GISTs) cited earlier work, identifying potential biomarkers predictive of histologic subtype (Nielsen, Hsu et al. 2003; West, Corless et al. 2004). In a more recent TMA study, markers including Ki-67, beta-catenin, CD44, and P- glycoprotein were empirically chosen based on previous literature and their protein expression examined on the arrayed specimens (Engellau, Bendahl et al. 2005).

~~383~~

~~Appendices~~

~~Immunostaining for Rb pathway proteins was similarly carried out by Sabah et al to elucidate the expression of cell cycle regulators in STS (Sabah, Cummins et al. 2006).~~

Ganti et al investigated EGFR and ErbB2 expression in a TMA comprising alveolar and embryonal rhabdomyosarcomas (ARMS and ERMS) (Ganti, Skapek et al. 2006). The decision to evaluate these markers was based on earlier work demonstrating EGFR expression in RMS cell lines (Ricci, Landuzzi et al. 2000) and a small Phase 1 trial of gefitinib in paediatric solid tumours refractory to other treatment (Daw, Furman et al. 2005).

Aberrant Wnt-β-catenin signaling and overexpression of Cyclin D1 in SS has also been reported in the literature (Saito, Oda et al. 2000; Saito, Oda et al. 2001; Ng, Gown et al. 2004). Horvai et al evaluated the correlation between β-catenin and Cyclin D1 expression in on an array incorporating 82 SS but noted no difference in expression between primary and metastatic tumours (Horvai, Kramer et al. 2006). Another member of the Wnt signaling pathway, Transducin Enhancer of Split 1 (TLE1) was overexpressed in SS (Terry, Saito et al. 2007), confirming findings in previous studies (Ng, Gown et al. 2004; Pretto, Barco et al. 2006).

~~IV.3 Limitations of Tissue Microarray and Counter Measures~~

~~IV.3.1 Sample size~~

Soft tissue sarcoma is a relatively rare malignancy, compared to epithelial cancers such as breast and colon cancer. In addition, over 50 histologic subtypes of STS exist. Despite the potential of TMA providing a high throughput modality for the investigation of biomarker expression on hundreds of samples, in practice, collecting archival specimens from a single institution yields small numbers. A collaborative approach with involvement of multiple centres is required to fully utilise the potential of tissue microarrays as a high throughput tool. This study was limited in its sample size by availability of archival specimens.

~~384~~

~~Appendices~~

~~IV.3.2 Tumour heterogeneity and TMA analysis~~

The validity of TMA in taking small cores from a large heterogeneous tumour specimen has been discussed in this thesis (Section 4.1.2) and has been the main focus of some earlier validation studies (Table 4.1). All of the previous validation studies were able to prove that two to three 0.6 mm cores yielded >90% concordance with corresponding conventional histologic slides (Camp, Charette et al. 2000; Gillett, Springall et al. 2000; Hoos, Urist et al. 2001). There was little extra information to be gained by increasing the sampling size to 2-4 mm cores. Sampling multiple cores from different regions was more representative.

TMA is a screening or large-scale research tool designed to investigate the prevalence of a selected molecular marker in the study population. It does not replace conventional histology in diagnosing individual patients, which requires detailed examination of factors such as excision margins, vascular or lymphatic invasion. Based on the above, 1 mm triplicate cores were used in this study. In view of the heterogeneous nature of soft tissue sarcomas, areas most representative of the overall tumour and its associated grade were selected for sampling. In addition, each TMA master block included at least one tumour that was sampled from more than one area, as a form of internal control. There was only one case in which there were discordant scores in terms of positive versus negative staining for pSTAT3.

While for the majority of cases, where expression of the biomarker being studied manifests diffuse expression, TMA can confidently be utilised, problems may arise in the investigation of markers which show very focal expression. In this thesis, the expression of pSTAT3 was found overall, to be low. In specimens displaying immunopositivity, the expression pattern was often very focal. In this case, there is a possibility of underestimating positivity where small cores rather than whole sections are used. Therefore, the expression of this particular marker may need to be more fully investigated using conventional whole tumour sections.

~~IV.3.3 Integrity of Cores~~

~~Tissue loss during sectioning and vigorous immunostaining procedures has been noted in most published validation studies on TMA (Schraml, Kononen et al. 1999; Camp,~~

~~385~~

~~Appendices~~

Charette et al. 2000; Gillett, Springall et al. 2000; Hoos, Urist et al. 2001; Dolled- Filhart, Camp et al. 2003). The proportion of “uninterpretable” cores in these studies has ranged from 10-30%. Ideally, uniformly long tissue cores should be taken from all donor blocks. In reality, there are always some thin donor blocks, either resulting from extensive prior sectioning or due to the thin tumour tissue itself. In some cases, this could be overcome by taking more than one donor core and placing it in tandem onto the same recipient hole. However, in one or two of the cases, the area representative of the appropriate grade of malignancy was too small to allow this without causing undue damage to the donor block.

However, in all, greater than 90% of the tumours arrayed in this study had cores available for analysis. This figure exceeds that in the literature of 75-85% “interpretable” spots. This improved rate is due to the re-sampling of tumours found to be “missing” on initial sectioning and H&E staining. Prior to re-sampling, 82% of tumours had interpretable cores, in line with published studies. Re-sampling clearly improved this rate. Many sarcomas grow to a very large size and some of those used in the present study exceeded 25 cm in diameter. This afforded the relative luxury of re- sampling. As in other studies, uninterpretable cores were a result of either loss of tissue on the TMA on arraying, sectioning or floating off during the immunostaining procedure (Schraml, Kononen et al. 1999; Dolled-Filhart, Camp et al. 2003).

~~IV.3.4 Scoring of Immunostaining~~

~~IV.3.4.1 Standardisation~~

There is generally a lack of standardisation with various scoring methods reported in different studies. Usually, a combination of extent (percentage of cells stained) and intensity of staining is employed. In one study, for instance, where EGFR staining was assessed, The percentage of positive tumor cells per slide (0% to 100%) was multiplied by the dominant intensity pattern of staining (1, negative or trace; 2, weak; 3, moderate; 4, intense); therefore, the overall score ranged from 0 to 400. However, this score was again reduced to a semi-quantitative score with 0 to 200 deemed to signify low expression, 201 to 300 intermediate and 301 to 400, high levels of expression (Hirsch, Varella-Garcia et al. 2003). Similar methods of scoring were employed by others for

~~386~~

~~Appendices~~

breast cancer and EGFR expression in paediatric rhabdomyosarcoma, eventually reducing the numeric score to a semi-quantitative score (Ginestier, Charafe-Jauffret et al. 2002; Ganti, Skapek et al. 2006).

Others have scored immunoreactivity as “strong” (diffuse and/or intense positive- staining in at least 20% of the cells), “weak” (any lesser degree of staining), negative, or uninterpretable where insufficient tumor cells were present within the TMA core sampled (Nielsen, Hsu et al. 2003). Where duplicate cores gave discordant results, the higher score was used. It has been shown that duplicate cores in tissue microarrays correct for focal expression of antigen in the majority of cases (Camp, Charette et al. 2000).

In another study, p-ERK, p-Akt, and p-STAT3 were quantitated subjectively as follows: negative = <5% of cells staining; + = diffuse but weak nuclear staining of cells with more than 5% of cells staining; and 2+ = diffuse and strong nuclear staining of cells with more than 5% of cells staining (Mukohara, Kudoh et al. 2003). In contrast, pSTAT3 scoring in one breast cancer TMA did not include an area variable at all, the justification of this being the small size of the cores (0.6 mm) (Dolled-Filhart, Camp et al. 2003). The staining intensity was graded on the following scale: 0, no staining; 1, weak staining; 2, moderate staining; and 3, intense staining.

Reproducibilty is therefore the issue where scoring of aTMA is carried out manually, on a semi-quantitative basis, as is usually the case. Automation in scoring for some markers is being considered and this may result in greater objectivity and reproducibility (Camp, Chung et al. 2002).

~~IV.3.4.2 Choice of Antibody~~

EGFR staining patterns and hence scoring can also be dependant on the antibody used. It has been shown that different proportions of STS showed different intensity of positive expression of EGFR depending on the antibody used, for each of the main histologic subtypes tested (Kersting, Packeisen et al. 2006). Only one of the antibodies was found on multivariate analysis to correlate with clinical outcome. All five

~~387~~

~~Appendices~~

~~antibodies tested, however, were specific for the extracellular domain of the EGF receptor. In other words, activated or phorphorylated EGFR was not investigated.~~

In the present thesis, the expression of EGFR, as assessed with the antibody used (Clone31G7, Zymed, CA, USA), was not found to correlate with prognosis. This concurs with the results reported by Kersting et al. Activated EGFR, however, was found to correlate with prognosis and this provides a more functional insight into the role of EGFR in tumour progression in STS.

~~IV.3.4.3 Commercial availability of Antibody~~

The availability of antibodies specific for the candidate marker can be an issue. Markers newly discovered on gene expression arrays may not as yet have the corresponding antibody for detecting its protein product expression. In these cases, the specific antibodies have to be developed before immunohistochemistry can be carried out on a large number of samples on a TMA.

~~IV.4 Clinical data (patients): Sydney Sarcoma Clinic Database~~

A prospectively kept database was maintained at the divisions of surgical oncology and radiotherapy in Prince of Wales Hospital, which recorded all clinical information and follow up data. This included clinical details at the time of first presentation to Prince of Wales Hospital (POWH), tumour pathology and treatment. If the initial presentation and surgery for soft tissue sarcoma had occurred at a peripheral hospital, this was also recorded. All subsequent presentations for local recurrence and distant metastases were also recorded. Local recurrence (LR) was defined as the first recurrence of tumour at the site of the primary tumour. Malignant lesions distant from the primary tumour identified synchronously or metachronously were defined as distant metastases. All neoadjuvant or adjuvant therapy was recorded. Follow up was maintained through the Sydney Sarcoma Clinic, a multidisciplinary clinic at POWH attended by surgeons, radiation and medical oncologists. Where follow up at the clinic was no longer required or the patient had ceased to attend, letters were sent to general practitioners, requesting latest known follow up data. Length of follow up was defined as the time from initial diagnosis to the date at which the patient was last seen or to date of death.

~~388~~

~~Appendices~~

~~IV.5 Construction of TMA master blocks~~

~~IV.5.1 The Equipment~~

The ATA100™ Advanced Tissue Arrayer (CHEMICON International, Inc., Temecula, CA) was used to construct the master or recipient blocks. The instrument comprised an integrated microscope, illuminated reference slide stage, donor block station, and recipient block station. There are matched needles for coring the recipient paraffin block and obtaining tissue cores from the donor blocks, the height of which were adjusted to suit the depth of each paraffin block. The reference slide and the donor block move together on a single platform, so that the area to be sampled can be identified on the reference slide under the microscope, then confirmed on the aligned donor block.

~~IV.5.2 Method~~

The array was constructed by first coring a hole in the recipient paraffin block. The area on the donor block to be sampled was confirmed on the corresponding slide. The core was then taken at the donor punch station, and delivered into the hole already made in the recipient block. This process was repeated for each core until the recipient master block or array was complete. Adjustments were made for needle height for each new donor and recipient block, according to the thickness of the block.

~~IV.5.3 Design of the arrays~~

1mm cores in triplicate for each tumour were taken, in a 10X10 format/template, allowing a maximum of 100 cores per master block, with sufficient space between cores (2mm), for ease of analysis. An orientating core was also placed at the top left hand corner of each block, from one of the donor blocks. In each of the master blocks A-D, additional cores were taken from randomly selected tumours (6-7 cores in total from 5 patients) from different regions of the tumour as internal controls for intratumour variability in each of the broad histologic subtypes. The design of each recipient block was asymmetric, leaving intentional spaces at certain rows or columns for ease of identification and orientation. Block A comprised leiomyosarcomas (LMS), Block B, liposarcomas (LPS), Block C, malignant fibrous histiocytomas (MFH) or pleomorphic sarcomas (PMS) and on Block D were arrayed the

~~389~~

~~Appendices~~

less common STS such as malignant peripheral nerve sheath tumours (MPNST), synovial sarcomas (SS), angiosarcomas (AS) and epithelioid sarcoma (EMS). The templates for each block were printed out in grid format to aid the construction of the master blocks. Blocks E and F were subsequently constructed, by sampling the tumours from A and B and C and D respectively that were found on H&E staining to have one or more “missing” cores (see below).

~~IV.5.4 Assessment of integrity of tissue cores on TMA blocks~~

~~IV.6 Immunostaining Protocols~~

~~IV.6.1 Optimisation~~

A series of optimisation experiments were initially carried out, using full sections of soft tissue sarcomas and positive controls. The parameters optimised included antigen retrieval methods, primary antibody dilution (1:20, 1:50 and 1:100) and incubation periods of primary antibodies. The optimised method used for each antibody is summarised in Table 4.4.

The initial positive controls employed for optimisation experiments were either tissues suggested in the product information for the given antibody, or those used in previous publications where the antibody had been used. Normal prostate and breast cancer tissue were used as positive controls for EGFR (Clone 31G7, Zymed Laboratories, Inc., CA, USA), while colon cancer specimens were used for activated EGFR (MAB3052, Chemicon International, Inc., Temecula, CA). Colon cancer was also used as the positive control for p42/44 MAPK, pSTAT3, while prostate cancer sections were used as positive controls for pAkt (Cell Signaling Technology, Inc., Danvers, MA, USA). Sections obtained from donor blocks of high and low grade soft tissue sarcomas were used as the samples in these optimisation experiments.

~~390~~

~~Appendices~~

Pepsin digestion for 15 min at 37 °C was found to be the optimal antigen retrieval method for EGFR (Zymed), while microwaving in citrate buffer was used for activated EGFR (Chemicon), pAkt and p44/42MAPK. Microwaving in EDTA in tris buffered saline tween (TBST) was carried out for pSTAT3.

~~IV.6.2 Immunostaining~~

~~All experiments were standardised to overnight incubation at 4 °C for the primary antibodies, as this provided the best immunostaining results for the antibodies.~~

The tissue microarray slides were warmed in an oven for a minimum of 2 h or overnight, deparaffinised with xylene rinses and then transferred through two changes of 100 % ethanol. Endogenous peroxidase activity was blocked by a10-15 min incubation in a 0.3 % hydrogen peroxide.

Antigen retrieval for all antibodies except for EGFR (Zymed) was performed by microwaving the slides in a sodium citrate buffer (pH 6.0) or EDTA/TBST for 4 min at high power, then 15 min at low power. The slides were then cooled for 20-30 min at room temperature. No cooling was required for pSTAT3. Pepsin digestion for 15 min at 37 °C was the optimised mode of antigen retrieval for EGFR (Zymed). Proteinase K, which had been described in previous studies using this antibody was found to be suboptimal (Nielsen, Hsu et al. 2003; Psyrri, Kassar et al. 2005). A 1 % pepsin stock solution was pre-prepared by adding 100 mg pepsin (Sigma-Aldrich, Buchs,Switzerland) to 10mM HCl (pH 2) and stored at − 20 °C. A 0.5 % working solution of pepsin was made up just prior to use.

After antigen retrieval, the slides were blocked with 2 % BSA/1x PBS (EGFR antibodies) or 5 % goat serum (phosphorylated downstream antibodies) for up to 1 h at room temperature to reduce nonspecific background staining. Primary antibody was applied and incubated overnight at 4 °C, flat in a humidified chamber, at the dilutions shown in Table 4.4. After a series of washes in PBS or TBS, the appropriate biotinylated secondary antibody (Vector Laboratories, Burlingame, CA) was applied. The slides were rinsed, and the secondary antibodies detected using the ABC avidin/biotin method using ABC Vectastain® (Vector Laboratories, Burlingame, CA),

~~391~~

~~Appendices~~

then visualised with a 10 min incubation of the enzyme or chromogen, liquid 3,3'- diaminobenzidine (DAB) in buffered substrate. Finally, the slides were counterstained with hematoxylin and mounted with Eukitt mounting medium.

~~392~~

~~Appendices~~

~~IV.7 Patient Characteristics **','!-. 2&-*-%'!" 2 5 10#!-0"#"',2",#7 0!-+ 2 1# ,"'1 .0#1#,2#"T~~

Table IV.1 Clinicopathologic Characteristics of Cohort + 0' *#, #6 *#VV "#+ *#VY "%#7 # ,WSTY #"' ,WT -**-5!.+2&1 #"' ,UU 8*'4#XS 8*'4#Q"'1# 1#$0##WT# 8*'4#Q5'2&"'12 ,2"'1# 1#Y& 8*'4#Q5'2&*-! *"'1# 1#S # 2&$0-+3T[ /-122-$-**-53.T 2'12-*-%7 /SY /.TT "% .TZ X 62�SZ 3+-300 "# STV TZ UVY #2 12 1'1SR >,),-5,T 2 %# SV UY TX :SV 3+-30 '8#!+ ≤WTZ 1W ,"≤SRUR 1SR UU # 0%',1++' **# 01SRW 0%', *1S ,"≤SR TR **-1#0SSU ,4-*4#"V[ '2# -620#+'27VR 303,)SX %# " ,"<#!)Y 8 "-+#, .#*4'1TU 3&-0 !'!* 4'27W )#.2& 3.#0$'!' *SZ ##.YU 'V3,),-5,R#6$2#. 2'#,21Q[5#0#* 121##,.0'-02--0',TRRVR&6$2#. 2'#,21QT5#0# * 121##,.0'-02--0',TRRU

~~393~~

~~Appendices~~

~~IV.8 Raw Data for Immunostaining of TMAs~~

~~IV.8.1 EGFR and Tumour Grade~~

Table IV.2 EGFR immunostaining in STS samples across Histologic Grade -&"( ,#% 2'4# .-1'2'4# 3-2 * %'12-*-%'! *-5%0 "# SS SV TW &0 "# ',2#0+#"' 2#%0 "# S Y Z &'%&%0 "# X VR VX +#2 12 2'!%0 "# S [ SR 3-2 * S[ YR Z[

~~IV.8.2 Activated EGFR and Tumour Grade~~

~~Table IV.3 Activated EGFR in STS samples across Histologic Grade~~

~~8!2'4 2#"-&"( ,#% 2'4# .-1'2'4# 3-2 * %'12-*-%'! *-5%0 "# ST SU TW &0 "# ',2#0+#"' 2# S Y Z %0 "# &'%&%0 "# R VX VX +#2 12 2'!%0 "# R SS SS 3-2 * SU YY [R~~

~~IV.8.3 p44/42MAPK and Tumour Grade~~

Table IV.4 p44/42 MAPK (Erk1 & Erk2) Immunostaining in STS samples .VV VT8.7 ,#% 2'4# .-1'2'4# 3-2 * %'12-*-%'! *-5%0 "# Z SY TW &0 "# ',2#0+#"' 2# R Z Z %0 "# &'%&%0 "# S VW VX +#2 12 2'!%0 "# R SR SR 3-2 * [ ZR Z[

~~394~~

~~Appendices~~

~~IV.8.4 pAkt and Tumour Grade~~

Table IV.5 Phosphorylated Akt Expression in STS samples across Histologic Grade .8)2 ,#% 2'4# .-1'2'4# 3-2 * %'12-*-%'! *-5%0 "# [ SX TW &0 "# ',2#0+#"' 2#%0 "# S Y Z &'%&%0 "# R VW VW +#2 12 2'!%0 "# R SS SS 3-2 * SR Y[ Z[

~~IV.8.5 pSTAT3 and Tumour Grade~~

Table IV.6 Phosphorylated STAT3 in STS samples across Tumour Grade .383U ,#% 2'4# .-1'2'4# 3-2 * %'12-*-%'! *-5%0 "# S[ X TW &0 "# ',2#0+#"' 2#%0 "# Y S Z &'%&%0 "# UR SV VV +#2 12 2'!%0 "# Z T SR 3-2 * XV TU ZY

~~IV.9 Correlation of Disease-Free Survival with Clinical Variables~~

Chi square testing was carried out for all clinical variables to assess correlation with disease free survival. This was done for both the whole patient cohort as well as the restricted group who did not undergo neoadjuvant treatment. The statistically significant variables such as stage, tumour grade, site and size are shown in the following table.

~~395~~

~~Appendices~~

Table IV.7 Correlation of Disease Free Survival and Clinical Variables "** +.*#1 (#- "(34 ,26!*3"#" + 0' *# + *3# . + *3# . 2 %#SV%0-3.1 .# 01-,*&'V/3 0# SZTZRY RTRRR SWT[YV RTRRS /')#*'&--"( 2'- S[TWVZ RTRRR SXTTZS RTRRS /',# 0V 7V/',# 0811-!' 2'-, SXTZYR RTRRR SUTWX[ RTRRR <-$: *'"* 1#1 ZV YS 2 %#TT%0-3.1& .# 01-,*&'V/3 0# SYT[UX RTRRR SWTWRS RTRRR *-,2',3'27*-00#!2'-, SXTRZW RTRRR SUTWYW RTRRR /')#*'&--"( 2'- SZTVTX RTRRR SWTYY[ RTRRR "'1�1-6 !23#12 /',# 0V 7V/',# 0811-!' 2'-, SYTYTU RTRRR SWTTZU RTRRR <-$: *'"* 1#1 ZV YS 0 "#*-541 2� .# 01-,*&'V/3 0# SWTZWR RTRRR SRTVR[ RTRRS *-,2',3'27*-00#!2'-, SUT[WZ RTRRR ZTYXS RTRRU /')#*'&--"( 2'- SZTT[S RTRRR SST[XZ RTRRS "'1�1-6 !23#12 /',# 0V 7V/',# 0811-!' 2'-, SWTXW[ RTRRR SRTTXT RTRRS <-$: *'"* 1#1 ZU YS 0 "#V%-3.1 .# 01-,*&'V/3 0# SXTV[X RTRRS SRTYTW RTRSU /')#*'&--"( 2'- SZT[TZ RTRRR STTTXU RTRRY /',# 0V 7V/',# 0811-!' 2'-, SWTTVV RTRRR [TYYU RTRRT <-$: *'"* 1#1 ZU YS '2#W%0-3.1 .# 01-,*&'V/3 0# SRTVWU RTRUU ZTYRZ RTRX[ /')#*'&--"( 2'- SSTYXZ RTRS[ [TRZR RTRW[ /',# 0V 7V/',# 0811-!' 2'-, WTU[V RTRTR WTTYY RTRTT <-$: *'"* 1#1 ZW YT '2#T%0-3.1# .# 01-,*&'V/3 0# YTXSY RTRRX YTTWS RTRRY *-,2',3'27*-00#!2'-, XTUVR RTRST WTZZZ RTRSW /')#*'&--"( 2'- YTWZU RTRRX YTSSS RTRRZ "'1�1-6 !23#12 /',# 0V 7V/',# 0811-!' 2'-, YTWTY RTRRX YTSWS RTRRY <-$: *'"* 1#1 ZW YT '8#U%0-3.1 .# 01-,*&'V/3 0# TT[WZ RTTTZ VTV[Z RTSRX /')#*'&--"( 2'- TT[YZ RTTTX VTVZS RTSRX /',# 0V 7V/',# 0811-!' 2'-, TTZZ[ RTRZ[ UT[VS RTRVY <-$: *'"* 1#1 ZV YS

'8#Y%0-3.1S4[!+Q7[!+-04WV!+Q5WV!+ #.2& ,"##1#!2'-, 0%',15#0#,-21'%,'$'! ,2 ," 0#,-21&-5, #'2#X%0-3.1S620#+'27Q03,)Q# "2#!)SW%.Q "- #*4'1&-0 6SX," &2 %#XX%0-3.1S2 %#1WQX',-,#%0-3. ,"YQZ',-2�%0-3.

~~396~~

~~Appendices~~

~~"() 9+~~

~~V. PROTOCOLS and ADDITIONAL CLINICAL DATA for ESTABLISHMENT and CHARACTERISATION of CELL LINES~~

~~397~~

~~Appendices~~

~~V.1 Collection of tissue~~

Informed consent was obtained prior to surgery from the patients. Once the specimen was removed, a small amount of tissue was taken from a viable portion of the tumour, avoiding areas of obvious necrosis. Selection of appropriate tissue was determined either in consultation with the surgeon performing the resection, or with the pathologist. A portion of the sample was placed in cryovials and snap frozen in liquid nitrogen as part of the tumour bank held within our unit. Every effort was made to minimise the time between resection of the tumour and snap freezing the tissue, to ensure preservation of RNA integrity. If sufficient tissue was available, an additional piece was snap frozen in OCT for sectioning and staining in the laboratory. Tissue for primary was collected in falcon tubes containing phosphate-buffered saline and transported to the laboratory. All patients had clinical data prospectively collected for our Sydney Sarcoma Unit database.

~~V.2 Primary culture and establishment of cell lines~~

Fresh tissue pieces were minced with scissors. Two or three 1mm3 explants were seeded directly to 100 mm2 tissue culture dishes that were cross scratched to facilitate ο attachment and media added. The dishes were placed at 37 C in a 5% CO2 incubator. For the early cultures, trypsin was used for enzymatic digestion. However, the yield of cells using this method was poor and for later cultures, collagenase was used instead. The majority of the minced tissue was disaggregated in Collagenase I at 100−200 IU/ml ο (Sigma, St Louis, MO) overnight at 37 C with agitation. If there was little fresh tissue available for culture, the suspension was examined after 4-6 h of digestion and the digestion reaction terminated once a cloudy cell suspension was seen. Following digestion, the enzyme action was stopped by placing the tubes on ice. The cell suspension was centrifuged and the pellet resuspended in Roswell Park Memorial Institute (RPMI)1640 medium supplemented with 10% foetal calf serum (FCS ; Cell Culture Laboratories, Cleveland, OH), penicillin G (100 U/ml), streptomycin (100 U/ml) and gentamicin (GIBCO BRL, Invitrogen, Carlsbad, CA) and seeded in 75 cm2 culture flasks (Falcon, BD Biosciences, Franklin Lakes, CA). The flasks were placed at ο 37 C in a 5% CO2 incubator and left undisturbed for the first 2-3 d. Media was changed every 3-4 d or weekly for the slower growing cultures. Subsequent subculturing of semi

~~398~~

~~Appendices~~

confluent cells was carried out using a 0.1% trypsin solution and cells resuspended in medium at a 1:2 dilution. After the first subculture, gentamicin was omitted from the media provided there was no microbial contamination. If required, enrichment of neoplastic cell populations in culture was attempted by quick trypsinisation. This involved marking areas of neoplastic cells, as distinct from fibroblasts on the culture flasks before adding trypsin. The flasks were watched carefully by light microscopy as cells detached. The cell suspension was transferred to a Falcon tube for centrifugation as soon as the malignant cells detached, leaving the more adherent fibroblasts still attached to the original culture flask. For some cell lines, this process was carried out for two to three subcultures. For the cell lines that were successfully subcultured for a number of passages without senescing, cell proliferation rate was determined using the Dimethylthiazoylyldiphenyltetrazolium bromide (MTT) assay (Sigma-Aldrich, Castle Hill, Sydney, NSW) (Appendix I).

~~V.3 Protocols for characterisation of cell lines~~

~~V.3.1 Morphology~~

Cells were examined in their culture flask using a phase contrast microscope (Zeiss IM35) at each passage for any changes in morphology or growth characteristics. Cells were also cultured in 4 well chamber slides (BD Biosciences, Franklin Lakes, CA), 5 seeded at 2.5 – 3 × 10 cells/well, and placed in the incubator at 37°C and 5% CO2 for 48 h to allow the cells to adhere. In preparation for staining, the media and chamber were removed and the slide washed with phosphate buffered saline (PBS). Cold acetone was used to fix the slides, before staining with haematoxylin and eosin (H&E).

~~V.3.2 Immunohistochemistry~~

~~Cells were harvested and seeded to 4 well chamber slides at 2.5 – 3 × 105 cells per well.~~

~~They were then allowed 48 h to adhere at 37°C in a 5% CO2 incubator before fixing with cold acetone as above and immunostaining. Plasma-thrombin clots of the cultured~~

~~399~~

~~Appendices~~

cells embedded in paraffin were also made by resuspending the harvested cell pellets in 50 μl plasma and 20 μl thrombin per 150 cm2 flask of cultured cells, then fixing in formalin and paraffin embedding as for other tissue. Staining for all markers other than p53, KAI1 (Dong, Lamb et al. 1995; Ow, Delprado et al. 2000) and KAI-1 COOH terminal interacting tetraspanin (Kitenin) (Lee, Park et al. 2004) was carried out using the alkaline phosphatase technique with fuschin as the chromogen. The markers mentioned were stained using the immunoperoxidase method. Dewaxing with HistoChoice and rehydration with ethanol was carried out for paraffin embedded tissue, while cold acetone fixation was performed for chamber slide cultures. In brief, the alkaline phosphate method involved incubating the slides with the primary antibody for 1 h, applying the link (biotinylated anti-mouse or anti-rabbit) and incubating again for 20 min. The slides were washed before applying the streptavidin alkaline phosphate conjugate and re-incubating for 20 min and washing again. The chromogen was applied for 5 min, then rinsed and the nuclei counterstained with Lille- Myers Haematoxylin. Antigen retrieval was required for p53 staining, using 500 ml 0.01M Citrate buffer (pH 6) at 1000 V for 5 min, then at one sixth power for 15 min followed by 15 min of cooling at room temperature (RT). The remainder of the indirect immunoperoxidase staining method was the same for the p53, KAI-1 and Kitenin antibodies. The slides were washed and endogenous peroxidase quenched using 0.3% H2O2. 3% (w/v) BSA was used to block non specific proteins before applying the primary antibodies. Incubation of p53 antibodies was performed overnight at 4°C in a humidifier box, whereas for KAI-1 and Kitenin, a 60 min incubation period at RT was sufficient. The slides were washed before applying the biotinylated horseradish peroxidase conjugated secondary antibodies for 30 min. A goat anti-mouse secondary antibody (Vector, Burlingame, CA) at 1:200 was used for p53 and goat anti-rabbit (Vector, Burlingame, CA) at 1:200 was used for KAI-1 and Kitenin. Harris haematoxylin was used for nuclear counterstaining. The primary antibodies used are summarised in Chapter 5, Table 5.1.

~~400~~

~~Appendices~~

~~V.3.3 Mutational Analysis~~

~~Genomic DNA was extracted from the cells using a commercially available kit (Qiagen) as per manufacturer’s instructions and 1 μg of DNA was used as the template.~~

~~V.3.3.1 TP53 mutation analysis for LMS-LFS cell line~~

Polymerase chain reaction (PCR) was used to amplify the region across exons 5 to 6 of the TP53 gene on chromosome 17p13.1 using the forward primer 5'-ccg tct tcc agt tgc ttt at-3' and the reverse primer 5'-tta acc cct cct ccc aga-3'. The DNA was amplified using the following cycle parameters: Denaturation at 95°C for 10 min, followed by 30 cycles of 95°C for 1 min, 61°C for 30 s and 72°C for 1 min, then a final extension step of 72°C for 5 min. Sequencing of the purified product was then carried out using the BigDye terminator reaction (Applied Biosystems) version 3 as per the manufacturer’s instructions.

~~V.3.3.2 KIT and PDGFRA mutation analysis for GIST-M cell line~~

The KIT mutation analysis was performed on both the paraffin embedded tumour specimen and the cell line. The paraffin block was sectioned (5 μM sections) onto uncoated slides and deparaffinised with Histoclear and rehydrated with ethanol. Methyl Green (0.1 − 1%) staining was then performed. Microdissection to select an appropriate area of malignant tissue was undertaken by a pathologist and the dissected tissue placed in 100 μl of lysis buffer (Qiagen, Doncaster, VIC). Proteinase K (Promega, Madison, WI) digestion was carried out at 56°C overnight.

~~Exons 9, 11, 13 and 17 of the KIT gene on chromosome 4q11-q12 were examined for mutations as was exon 18 of PDGFRA on 4q11-q12. The primers for these are given in Chapter 5, Table 5.2~~

The PCR was set up using AmpliTaq Gold™ DNA polymerase with 2 mM MgCl2. The cycling parameters were as follows: Denaturation at 94°C for 10 min, followed by 45 cycles of 94°C for 45 s, 55°C for 30 s and 72°C for 30 s, then a final extension step at 72°C for 10 min. The PCR products were electrophoresed on a 2% agarose gel to confirm the bands, then purified using Exonuclease I and Shrimp Alkaline Phosphatase

~~401~~

~~Appendices~~

(SAP). As above, sequencing of the purified product was carried out using the BigDye terminator. The thermal cycler was set to ramp up rapidly to 96°C and hold for a minute, followed by 25 cycles of 96°C for 10 s, 50°C for 5 s and 60°C for 4 min. The sequenced products were purified and analysed by capillary electrophoresis on the ABI3100 sequencer. Incorporation of the fluorescence labeled ddNTP during the sequencing reaction allows the separated DNA fragments to be visualised on the sequencer.

~~V.3.4 Analysis of Telomere Maintenance Mechanism (TMM)~~

~~V.3.4.1 TRAP Assay~~

Protein was extracted from cells using ice cold lysis buffer as described by Kim et al in 1994 (Kim, Piatyszek et al. 1994) and stored in 1 μg/μl aliquots for the TRAP assay. Primers used for the PCR were as given below: Primer M2 (or TS) 5’-AAT CCG TCG AGC AGA GTT-3’ Primer CX 5’- CCC TTA CCC TTA CCC TTA CCC TAA-3’ PCR was carried out with the primers, Taq polymerase dNTPs and TRAP buffer for 30 cycles of denaturing at 94°C for 10 s, annealing at 50°C for 25 s and extension at 72°C for 30 s, followed by one cycle of 94°C for 15 s, annealing at 50°C for 25 s and extension at 72°C for one min. The product was then run on a 10 % non-denaturing polyacrylamide gel and stained with SYBR Green (Molecular Probes) for visualisation. Lysis buffer was used as the negative control and the osteosarcoma cell line MG-63, which is known to exhibit telomerase activity, was the positive control. Heat inactivation was used on the positive samples to confirm presence of telomerase. Negative samples were spiked with the positive control to exclude the possibility of presence of telomerase inhibitors.

~~V.3.4.2 ALT associated PML bodies (APB) staining~~

Cells were cultured on chamber slides, washed and fixed before incubating with anti- PML rabbit antibody (Chemicon, Temecula, CA), followed by anti-rabbit FITC goat antibody (Sigma, St Louis, MO). Hybridisation with the Cy3 labeled peptide nucleic acid FISH probe for telomeric DNA (Applied Biosystems, Framingham, MA) was then carried out for three hours before counterstaining nuclei with 4′6-Diamidino-2-

~~402~~

~~Appendices~~

~~phenylindole (DAPI) (Sigma). Colocalisation of PML bodies and telomeric DNA within the nuclei of the cells represents presence of APBs.~~

~~V.4 Patients and Source Tumours~~

~~V.4.1 DOS5~~

The 40 year old patient had complained of weight gain and specifically, increased abdominal girth and an epigastric mass. A large 30 cm retroperitoneal tumour was seen on computed tomography (CT) scanning and he underwent surgical resection of this in November 2003. The tumour was a large circumscribed fatty mass approximately 40 cm at its largest diameter, with a thin fibrous capsule. The histologic appearance was that of a well differentiated liposarcoma, with only scant lipoblasts. The majority of the tumour had the appearance of a lipoma. The capsule was incised and tissue was obtained from within the bulk of the homogenous fatty tumour for culture. The tissue was minced with a sterile scalpel and scissors and placed on cross scratched tissue culture dishes with media supplemented with antibiotics.

~~V.4.2 DOS6~~

The patient in this case had initially presented at another hospital with a discharging buttock mass that was debrided and biopsied. A subsequent wound infection required further washout and packing of the wound. At this juncture, a CT scan revealed a large underlying soft tissue mass that had not been present on the previous films. The definitive surgical procedure carried out at our institution involved a wide local excision with skeletonisation of the sciatic nerve. The tumour was found at the time to be abutting bone at its deep margin. The tumour was large, 20 cm in diameter, infiltrating skeletal muscle. There was a large amount of myxoid tissue and fat, with varying degrees of cellularity, moderate pleomorphism and frequent mitoses. A diagnosis of myxoid liposarcoma was made.

~~403~~

~~Appendices~~

Tissue was obtained from the excised operative specimen. A portion of the fragments were minced and seeded directly to tissue culture dishes and the remainder, digested with trypsin for 15 min at 37°C, before centrifuging, resuspending and seeding to flasks. The yield of cells from both of these methods was low, with a large amount of fat globules present rather than malignant cells. This was in keeping with the histopathology of the source tumour, which did have areas of hypocellularity.

~~V.4.3 DOS7~~

The 33 year old female patient had presented with a 12 month history of a right thigh mass, which on incision biopsy at a regional hospital had been reported to be a lipoma. The mass was clinically however thought to be a sarcoma and the patient underwent surgical resection at our institution in November 2003. The multilobular 14 cm mass was another well differentiated lipoma-like liposarcoma. Fragments of tissue from this tumour were directly seeded onto tissue culture dishes and flasks, without enzyme digestion, as this had not appeared to increase the yield of cells for the previous culture initiated from a well differentiated liposarcoma, DOS5.

~~V.4.4 DOS8~~

The 49 year old patient in this case had been referred from a regional hospital with a 3 month history of pain and an abdominal mass. A diagnostic biopsy had revealed a pleomorphic sarcoma in the retroperitoneum. The histology of the surgically resected specimen concurred, with the finding of a high grade pleomorphic sarcoma infiltrating the psoas muscle and right renal cortex. Tissue from this tumour was minced and placed directly into culture without enzymatic digestion.

~~V.4.5 DOS9~~

This patient had presented with a 12 month history of a mass in his right posterior thigh, approximately 5 cm in size. The core biopsy done at the referring hospital had been suggestive of a malignant fibrous histiocytoma (MFH). He underwent surgical resection of the tumour in January 2004. The definitive histopathologic diagnosis confirmed a grade 3 pleomorphic sarcoma showing strong staining for vimentin, weaker staining for

~~404~~

~~Appendices~~

desmin and cytokeratin AE1/3 and no staining for smooth muscle actin (SMA), S100 and HMB45. Tissue from this tumour was minced and placed directly into culture without enzymatic digestion. There was initial rapid outgrowth from the explanted tissue, allowing subculture to be carried out. However, the cell strain was cross contaminated by an aggressive colorectal carcinoma cell line and therefore not cultured further.

~~V.4.6 DOS10~~

The 47 year old patient underwent a right below knee amputation in Canberra for an anterior tibial compartment tumour. Microscopically both anterior and posterior compartments were involved. Initially, it was considered to be a soft tissue sarcoma invading the bone but this diagnosis was revised to that of a grade 3 osteogenic sarcoma with secondary involvement of the soft tissues. Surgery was performed in Canberra, with tissue collected and placed in culture there by our collaborators at ACT Pathology. A flask of cells in culture was received at our laboratory and the culture maintained for a number of passages until the diagnosis of an osteogenic sarcoma was confirmed. As this thesis is concerned with soft tissue sarcoma rather than osteogenic sarcoma, the cells were frozen down at this point and stored.

~~V.4.7 DOS11~~

This 18 year old patient had an inconclusive core biopsy at a referral centre of her 6 cm posterior thigh mass. A magnetic resonance image revealed a mass involving the right biceps femoris and she underwent an incision biopsy at our hospital. Tissue was obtained at this point for initiating the primary culture, before the patient had definitive surgery. The histopathologic report on the biopsy was that of a neurofibroma, however the definitve diagnosis after surgical resection was neurofibrosarcoma. Only a small amount of tissue was obtained and enzymatic digestion was not carried out. The tissue was explanted directly onto culture dishes.

~~405~~

~~Appendices~~

~~V.4.8 DOS12~~

This 48 year old gentleman had a tumour involving the left clavicular region, extending to the first rib and to the midline invading the manubrium. Microscopically, the tumour was found to be an undifferentiated high grade sarcoma with lymphovascular infiltration. The patient went on rapidly to develop vertebral metastases. In February 2004, tissue fragments obtained from this tumour were minced and the majority of it digested with Collagenase (100 IU/ml) at 37°C overnight. A small amount of the tissue was seeded directly onto dishes. The cells were however too sensitive to collagenase digestion and subculturing was carried out for the outgrowths from the directly seeded explants only.

~~V.4.9 DOS13~~

The 35 year old patient had presented with an 8 − 12 month history of a large right axillary mass. The diagnosis based on the preoperative core biopsy was that of a liposarcoma. The definitive histopathology of the resected tumour confirmed it to be a low grade myxoid liposarcoma. The capsule was incised and tissue obtained for primary culture. As for the previous culture, some of the tissue was digested in collagenase overnight and the remainder was placed directly onto cross scratched culture dishes. The flasks the containing collagenase treated cell suspension reached 80% confluence in six days and was passaged. With the subsequent subculture, RNA was extracted and frozen for future analysis.

~~V.4.10 DOS14~~

This patient had presented in December 2003 to a district hospital with a mass at the right shoulder, which was resected and found to be a follicular dendritic cell sarcoma. This rare subtype of soft tissue sarcoma, arising from antigen-presenting cells, is thought to behave as an intermediate grade tumour (Chan, Fletcher et al. 1997), with local recurrence occurring in 50% of patients and distant metastases in 25%. There were no adverse features in the primary tumour. The patient represented in April 2004 with an axillary mass and was referred to our institution. The axillary lymph nodes were involved with secondary tumour and were resected.

~~406~~

~~Appendices~~

Tissue for primary culture was obtained from one of the malignant nodes. As before, some of the tissue was subjected to overnight enzymatic digestion with collagenase, and the rest seeded directly to culture dishes.

~~V.4.11 DOS17~~

This 22 year old patient presented with a mass in the left clavicular region. A core biopsy revealed an embryonal rhabdomyosarcoma, which is usually more commonly seen in the paediatric population. As this is a chemosensitive tumour, the patient received neoadjuvant chemotherapy, with a clinical reduction in tumour size. Surgery was subsequently performed in August 2004 to remove the tumour. The fragments of tissue taken from the tumour were digested with collagenase by placing the tubes in a shaker at 37°C. One falcon tube had a cloudy suspension after 6 hours of digestion and was therefore seeded at this juncture. The remaining tube was left to digest in collagenase overnight.

~~V.4.12 DOS18~~

A core biopsy of this patient’s left hip mass had revealed a spindle cell tumour. The histology of the resected tumour showed a decubitus ulcer with fibrosis, fat necrosis and fibrinoid change consistent with ischaemic fasciitis, with no evidence of malignancy. Tissue sampled for primary culture was treated with collagenase overnight as above. There was no cloudy cell suspension after 6 hours of digestion. The culture was maintained for 4 passages and then frozen down as the source tumour was found to be benign.

~~V.4.13 DOS19~~

This gentleman underwent a surgical resection of his right thigh tumour in Canberra. A grade 2 myxoid liposarcoma was the histologic diagnosis. The primary culture was initiated at the collaborator’s laboratory in Canberra by directly seeding tissue fragments to culture flasks. No enzymatic digestion was carried out. A flask of cells was received in our laboratory approximately one month later. There was scant growth of

~~407~~

~~Appendices~~

~~cells even after a further month of culture. There were insufficient cells for subculturing or freezing.~~

~~V.4.14 DOS20~~

The 24 year old patient initially presented in March 2004 with a history of left sided abdominal pain, was investigated with a CT scan and found to have a retroperitoneal mass. She subsequently became pregnant and her symptoms subsided. She became symptomatic again post partum and the mass was found to have grown in the intervening period since the last CT scan. A preoperative core biopsy revealed malignant cells with a mitotic rate of 17/10 high power fields (HPF) and focal positivity for cytokeratin and SMA. Surgical resection was performed in February 2005. Both direct seeding of tissue fragments and collagenase digestion was carried out. For the latter, minced tissue in a falcon tube with collagenase (200 IU/ml) in RPMI was placed in a shaker at 37°C overnight. The cell suspension was centrifuged and the pellet resuspended in media and transferred to culture flasks, with a very high yield of cells that reached confluence within 2 − 3 days. Some flasks however, had microbial contamination and had to be discarded. The remainder of the flasks with collagenase treated cells apoptosed by the third subculture. The cells that were seeded directly without collagenase digestion have been maintained in continuous culture at the time of writing.

~~V.4.15 DOS21~~

The 21 year old patient presented with a chest wall mass. A CT scan delineated a mass involving the left lower ribs and lateral abdominal wall. Biopsies revealed histologic features consistent with a Ewing’s sarcoma but PAS and glycogen stains, which are characteristic of Ewing’s were negative. Chromosomal analysis also failed to identify the t(11;22) translocation associated with 95% of Ewing’s tumours. The operation performed in February 2005 involved resection of the tumour, including the 10th − 12th ribs, followed by reconstruction of the lateral chest and abdominal wall. The tumour microscopically involved the 11th rib and the diaphragm and was composed of cells with oval nuclei, inconspicuous nucleoli and little cytoplasm. The features were again

~~408~~

~~Appendices~~

consistent with an extraskeletal Ewing’s sarcoma / primitive neuroectodermal tumour (PNET). Again, for the primary culture, both enzymatic digestion and direct seeding of tissue fragments was carried out. The outgrowth form the explants were subcultured and quick trypsinisation was performed to enrich the malignant cell populations. These cells were still in culture at the time of writing and could be characterised fully if they survive further passaging.

~~V.4.16 DOS23~~

The diagnostic core biopsy taken from this patient’s left posterior thigh mass was reported as a malignant triton tumour. This is thought to be a malignant peripheral nerve sheath tumour (MPNST) with rhabdomyoblastic differentiation. The patient had a wide local excision performed of the tumour which involved the biceps femoris muscle. The lobulated tumour was comprised of pleomorphic spindle cells, with frequent mitoses and areas of necrosis. The malignant cells were negative for most markers on immunohistochemistry and it was reported as a grade 3, undifferentiated spindle cell sarcoma. Some of the tissue fragments were minced and placed directly on culture dishes that had been cross scratched to facilitate adhesion. The remainder was subjected to overnight collagenase digestion, with a high yield of cells at the end of this period. The cells reached subconfluence and needed to be subcultured weekly, and were maintained in culture at the time of writing. Cells at passage 16 were karyotyped.

~~V.4.17 DOS24~~

On this admission the patient presented with a right sided chest wall mass. She had had previous resections of soft tissue sarcomas, for which she had also received adjuvant radiotherapy. The primary tumour had involved the paravertebral muscles. This chest wall metastasis had presented outside the field of her previous radiation. A core biopsy had confirmed it to be a pleomorphic sarcoma. It was widely excised in March 2005.

~~409~~

~~Appendices~~

Tissue was obtained as before, and cultured both with and without enzymatic digestion. The directly seeded tissue initially had very little outgrowth from explants. The collagenase treated tissue yielded large cell pellets that were resuspended in media and seeded to several flasks. There was initial rapid cell proliferation requiring subculturing. However, this was followed by apoptosis of the malignant cells and overgrowth of fibroblasts. Malignant cells appeared to be eventually restricted to the dishes that had the directly seeded explants. These were maintained in culture at the time of writing.

~~V.4.18 DOS25~~

The presenting complaint of this 45 year old patient was tenesmus. She was investigated with a colonoscopy which was normal, and then went on to have magnetic resonance imaging (MRI), which delineated an 8 cm mass attached to the levator ani. Surgical resection was performed in May 2005 and the histopathologic diagnosis was a grade 1 leiomyosarcoma (LMS). Positive markers on immunoperoxidase studies included desmin, SMA and bcl-2. The primary culture was initiated as for the previous specimens. The cells were still in culture at the time of writing.

~~V.5 Patient and Tumour Source for GIST-M (DOS15) cell line~~

~~V.5.1 Patient Presentation~~

The 62 year old patient presented in Canberra with a 6 month history of abdominal and back pain. A CT scan carried out at this stage revealed a large retroperitoneal solid and cystic mass. The complicating factor was that it appeared to involve her only remaining kidney and was therefore considered to be unresectable. She had had a right nephrectomy as a child. Core biopsies of the mass were undertaken which revealed features consistent with a gastrointestinal stromal tumour that stained positive for the marker ckit. She was therefore treated with the tyrosine kinase inhibitor imatinib mesylate (STI 571/Gleevec®/Glivec®, Novartis, Basel, Switzerland) for a period of 12 months. During this time, her symptoms did not subside and the tumour did not change in size

~~410~~

~~Appendices~~

radiologically. A positron emission tomography (PET) scan did, however, show a marked decrease in uptake of 18Fluorodeoxy-glucose (18FDG), indicating a response to the treatment (van Oosterom, Judson et al. 2001; Israel-Mardirosian and Adler 2003; Reddy, Reddy et al. 2003; Desai, Maki et al. 2004; Gayed, Vu et al. 2004) within 6 months of commencing imatinib. The patient was referred to our institution for consideration of surgical resection, in view of her ongoing symptoms. On review, it was thought that the kidney was not involved with tumour and as such, she underwent surgical resection of the malignancy in April 2004. Intraoperatively, the tumour appeared to arise from the antimesenteric border of the fourth part of the duodenum and the jejunum. The tumour was resected and the kidney preserved. Following surgery, the patient remained well on initial follow up but within a year was found to have recurrent disease involving the para aortic lymph nodes and nodes anterior to the left renal vein. She was treated again with imatinib at this stage and a repeat CT Scan showed decrease in the size of the nodes, while a repeat FDG-PET scan showed no active glucose avid foci of disease. The dose of imatinib was however increased empirically, due to persistant symptoms of pain.

~~V.5.2 Histology of Tumour~~

The resected tumour was well circumscribed and found histologically to arise from the muscularis externa of the duodenum, forming a solid and cystic serosal mass, with areas of haemorrhage and necrosis. It was comprised of pleomorphic spindle cells with a low (1/50 HPF) mitotic rate. The cells stained positive for c-kit (CD117), desmin and vimentin and negative for CD34, S100 and smooth muscle actin (SMA). It was designated a malignant GIST based on its size (>5 cm). Tissue was obtained at the time of surgical resection for establishing a primary culture.

~~411~~

~~Appendices~~

~~V.6 Patient and Tumour Source for LMS-LFS (DOS16) cell line~~

~~V.6.1 Patient Presentation and Histology of Tumour~~

The 37 year old patient was known to have Li-Fraumeni Syndrome (Chapter 5, Figure 5.16). She had had bilateral mastectomies for ductal carcinoma in situ (DCIS) and a frontal craniotomy for a Grade 2 Glioma. On this occasion she had experienced 6 months of vague back pain and more recently, episodes of abdominal pain for which she presented to our Emergency Department at Prince of Wales Hospital. A CT Scan confirmed a large 12 cm left sided retroperitoneal mass. Histopathologic examination of the core biopsy was unable to distinguish between a sarcoma and a carcinoma and the differential diagnosis therefore included adrenocortical carcinoma, renal cell carcinoma and sarcoma. The patient underwent a laparotomy in July 2004 for resection of the tumour. Histology of the resected specimen (Chapter 5, Figure 5.17) showed a high grade spindle cell sarcoma with features consistent with a leiomyosarcoma (LMS). The cells were arranged in poorly defined fascicles and had pleomorphic nuclei and numerous mitotic figures. There was extensive necrosis and the tumour invaded multiple adjacent organs. The tumour, as summarised on Table 2, stained positive for Vimentin, Desmin, CD10 and Cytokeratins (CAM5.2 and AE1/AE3), with patchy staining for CD99 and patchy staining for Smooth muscle actin (SMA). The tumour was negative for bcl2 and S100. A small amount of fresh tissue was obtained at surgery for establishing this cell line.

~~The patient received postoperative chemoradiotherapy but developed widespread metastatic disease over the ensuing months and succumbed 8 months after her resection.~~

~~V.6.2 Family Pedigree~~

The diagnosis of Li-Fraumeni syndrome in this family was made once the patient KS was investigated for the bilateral breast tumours and glioma. A family history was obtained (Chapter 5, Figure 5.16) and during the course of this investigation, the patient’s daughter, MS, presented with RMS requiring surgical resection. Mutational analysis was carried out on both KS and MS, examining exons five to eight by PCR

~~412~~

~~Appendices~~

amplification. A mutation was detected in the TP53 locus (chromosome17p13.1) at exon 6, codon 213, causing a change from CGA to TGA, a termination codon. This mutation, which has been reported to occur in LFS (Hainaut ; Frebourg, Barbier et al. 1995; Olivier, Eeles et al. 2002) results in the production of a truncated protein.

~~413~~

~~Bibliography~~

~~& &* "20~~

~~(1995). ISCN. ISCN 1995: an international system for human cytogenetic nomenclature. Basel, S. Karger.~~

~~(SMAC), S. M.-a. C. (1997). Adjuvant chemotherapy for localised resectable soft-tissue sarcoma of adults: meta-analysis of individual data. Lancet 350: 1647-1654.~~

Albanell, J., Rojo, F., Averbuch, S., Feyereislova, A., Mascaro, J. M., Herbst, R., LoRusso, P., Rischin, D., Sauleda, S., Gee, J., Nicholson, R. I. and Baselga, J. (2002). Pharmacodynamic Studies of the Epidermal Growth Factor Receptor Inhibitor ZD1839 in Skin From Cancer Patients: Histopathologic and Molecular Consequences of Receptor Inhibition. Journal of Clinical Oncology 20: 110-124.

~~Albritton, K. H. and Randall, R. L. (2005). Prospects for targeted therapy of synovial sarcoma. Journal of Pediatric Hematology/Oncology. 27: 219-222.~~

Alizadeh, A. A., Eisen, M. B., Davis, R. E., Ma, C., Lossos, I. S., Rosenwald, A., Boldrick, J. C., Sabet, H., Tran, T., Yu, X., Powell, J. I., Yang, L., Marti, G. E., Moore, T., Hudson, J., Jr., Lu, L., Lewis, D. B., Tibshirani, R., Sherlock, G., Chan, W. C., Greiner, T. C., Weisenburger, D. D., Armitage, J. O., Warnke, R., Levy, R., Wilson, W., Grever, M. R., Byrd, J. C., Botstein, D., Brown, P. O. and Staudt, L. M. (2000). Distinct types of diffuse large B-cell lymphoma identified by gene expression profiling. Nature 403: 503-511.

~~Alizadeh, A. A., Ross, D. T., Perou, C. M. and van de Rijn, M. (2001). Towards a novel classification of human malignancies based on gene expression patterns. Journal of Pathology 195: 41-52.~~

Allander, S. V., Illei, P. B., Chen, Y., Antonescu, C. R., Bittner, M., Ladanyi, M. and Meltzer, P. S. (2002). Expression profiling of synovial sarcoma by cDNA microarrays: association of ERBB2, IGFBP2, and ELF3 with epithelial differentiation. American Journal of Pathology. 161: 1587-1595.

Allander, S. V., Nupponen, N. N., Ringner, M., Hostetter, G., Maher, G. W., Goldberger, N., Chen, Y., Carpten, J., Elkahloun, A. G. and Meltzer, P. S. (2001). Gastrointestinal stromal tumors with KIT mutations exhibit a remarkably homogeneous gene expression profile. Cancer Research 61: 8624-8628.

~~Altieri, D. C. (2001). The molecular basis and potential role of survivin in cancer diagnosis and therapy. Trends in Molecular Medicine 7: 542-547.~~

~~Altman, A. and Villalba, M. (2003). Protein kinase C-theta (PKCtheta): it's all about location, location, location. Immunological Reviews 192: 53-63.~~

~~414~~

~~Bibliography~~

Andersson, J., Sjogren, H., Meis-Kindblom, J. M., Stenman, G., Aman, P. and Kindblom, L. G. (2002). The Complexity of KIT Gene Mutations and Chromosome Rearrangements and Their Clinical Correlation in Gastrointestinal Stromal (Pacemaker Cell) Tumors. American Journal of Pathology 160: 15-22.

Angelov, L., Salhia, B., Roncari, L., McMahon, G. and Guha, A. (1999). Inhibition of Angiogenesis by Blocking Activation of the Vascular Endothelial Growth Factor Receptor 2 Leads to Decreased Growth of Neurogenic Sarcomas. Cancer Research 59: 5536-5541.

Antonescu, C. R., Besmer, P., Guo, T., Arkun, K., Hom, G., Koryotowski, B., Leversha, M. A., Jeffrey, P. D., Desantis, D., Singer, S., Brennan, M. F., Maki, R. G. and DeMatteo, R. P. (2005). Acquired Resistance to Imatinib in Gastrointestinal Stromal Tumor Occurs Through Secondary Gene Mutation. Clinical Cancer Research 11: 4182- 4190.

Antonescu, C. R., Erlandson, R. A. and Huvos, A. G. (2000). Primary fibrosarcoma and malignant fibrous histiocytoma of bone--a comparative ultrastructural study: evidence of a spectrum of fibroblastic differentiation. Ultrastructural Pathology 24: 83-91.

Antonescu, C. R., Leung, D. H., Dudas, M., Ladanyi, M., Brennan, M., Woodruff, J. M. and Cordon-Cardo, C. (2000). Alterations of Cell Cycle Regulators in Localized Synovial Sarcoma : A Multifactorial Study with Prognostic Implications. American Journal of Pathology 156: 977-983.

Antonescu, C. R., Viale, A., Sarran, L., Tschernyavsky, S. J., Gonen, M., Segal, N. H., Maki, R. G., Socci, N. D., DeMatteo, R. P. and Besmer, P. (2004). Gene Expression in Gastrointestinal Stromal Tumors Is Distinguished by KIT Genotype and Anatomic Site. Clinical Cancer Research 10: 3282-3290.

Arnold, A., Kim, H. G., Gaz, R. D., Eddy, R. L., Fukushima, Y., Byers, M. G., Shows, T. B. and Kronenberg, H. M. (1989). Molecular cloning and chromosomal mapping of DNA rearranged with the parathyroid hormone gene in a parathyroid adenoma. Journal of Clinical Investigation 83: 2034-2040.

Aryee, D. N. T., Ambros, I., Ambros, P. F., Muehlbacher, K., Luegmayr, A., Yoshie, O. and Kovar, H. (2002). Frequent Low Level Expression in Ewing Sarcoma Family Tumors and Widespread Absence of the Metastasis Suppressor KAI1/CD82 in Neuroblastoma. Pediatric Research 52: 279-285.

Ashburner, M., Ball, C. A., Blake, J. A., Botstein, D., Butler, H., Cherry, J. M., Davis, A. P., Dolinski, K., Dwight, S. S., Eppig, J. T., Harris, M. A., Hill, D. P., Issel-Tarver, L., Kasarskis, A., Lewis, S., Matese, J. C., Richardson, J. E., Ringwald, M., Rubin, G. M. and Sherlock, G. (2000). Gene ontology: tool for the unification of biology. The Gene Ontology Consortium. Nature Genetics. 25: 25-9.

~~Baba, I., Shirasawa, S., Iwamoto, R., Okumura, K., Tsunoda, T., Nishioka, M., Fukuyama, K., Yamamoto, K., Mekada, E. and Sasazuki, T. (2000). Involvement of~~

~~415~~

~~Bibliography~~

~~Deregulated Epiregulin Expression in Tumorigenesis in Vivo through Activated Ki-Ras Signaling Pathway in Human Colon Cancer Cells. Cancer Research 60: 6886-6889.~~

Bai, C., Connolly, B., Metzker, M. L., Hilliard, C. A., Liu, X., Sandig, V., Soderman, A., Galloway, S. M., Liu, Q., Austin, C. P. and Caskey, C. T. (2000). Overexpression of M68/DcR3 in human gastrointestinal tract tumors independent of gene amplification and its location in a four-gene cluster. Proceedings of the National Academy of Sciences of the United States of America 97: 1230-1235.

Baird, K., Davis, S., Antonescu, C. R., Harper, U. L., Walker, R. L., Chen, Y., Glatfelter, A. A., Duray, P. H. and Meltzer, P. S. (2005). Gene expression profiling of human sarcomas: insights into sarcoma biology. Cancer Research 65: 9226-9235.

~~Bala, S. and Peltomaki, P. (2001). CYCLIN D1 as a genetic modifier in hereditary nonpolyposis colorectal cancer. Cancer Research 61: 6042-6045.~~

Baldini, E. H., Goldberg, J., Jenner, C., Manola, J. B., Demetri, G. D., Fletcher, C. D. and Singer, S. (1999). Long-term outcomes after function-sparing surgery without radiotherapy for soft tissue sarcoma of the extremities and trunk. Journal of Clinical Oncology. 17: 3252-9.

~~Ball, E. M., Mellor, S. L. and Risbridger, G. P. (2004). Cancer progression: is inhibin alpha from Venus or Mars? Cytokine Growth Factor Reviews 15: 291-296.~~

Bammler, T., Beyer, R. P., Bhattacharya, S., Boorman, G. A., Boyles, A., Bradford, B. U., Bumgarner, R. E., Bushel, P. R., Chaturvedi, K., Choi, D., Cunningham, M. L., Deng, S., Dressman, H. K., Fannin, R. D., Farin, F. M., Freedman, J. H., Fry, R. C., Harper, A., Humble, M. C., Hurban, P., Kavanagh, T. J., Kaufmann, W. K., Kerr, K. F., Jing, L., Lapidus, J. A., Lasarev, M. R., Li, J., Li, Y. J., Lobenhofer, E. K., Lu, X., Malek, R. L., Milton, S., Nagalla, S. R., O'Malley J, P., Palmer, V. S., Pattee, P., Paules, R. S., Perou, C. M., Phillips, K., Qin, L. X., Qiu, Y., Quigley, S. D., Rodland, M., Rusyn, I., Samson, L. D., Schwartz, D. A., Shi, Y., Shin, J. L., Sieber, S. O., Slifer, S., Speer, M. C., Spencer, P. S., Sproles, D. I., Swenberg, J. A., Suk, W. A., Sullivan, R. C., Tian, R., Tennant, R. W., Todd, S. A., Tucker, C. J., Van Houten, B., Weis, B. K., Xuan, S., Zarbl, H. and Members of the Toxicogenomics Research, C. (2005). Standardizing global gene expression analysis between laboratories and across platforms. Nature Methods 2: 351-356.

Bandyopadhyay, S., Pai, S. K., Hirota, S., Hosobe, S., Takano, Y., Saito, K., Piquemal, D., Commes, T., Watabe, M., Gross, S. C., Wang, Y., Ran, S. and Watabe, K. (2004). Role of the putative tumor metastasis suppressor gene Drg-1 in breast cancer progression. Oncogene. 23: 5675-5681.

Bandyopadhyay, S., Wang, Y., Zhan, R., Pai, S. K., Watabe, M., Iiizumi, M., Furuta, E., Mohinta, S., Liu, W., Hirota, S., Hosobe, S., Tsukada, T., Miura, K., Takano, Y., Saito, K., Commes, T., Piquemal, D., Hai, T. and Watabe, K. (2006). The tumor metastasis suppressor gene Drg-1 down-regulates the expression of activating transcription factor 3 in prostate cancer. Cancer Research 66: 11983-11990.

~~416~~

~~Bibliography~~

~~Bange, J., Zwick, E. and Ullrich, A. (2001). Molecular targets for breast cancer therapy and prevention. Nature Medicine 7: 548-552.~~

Barel, D., Avigad, S., Mor, C., Fogel, M., Cohen, I. J. and Zaizov, R. (1998). A novel germ-line mutation in the noncoding region of the p53 gene in a Li-Fraumeni family. Cancer Genetics & Cytogenetics 103: 1-6.

Barlund, M., Forozan, F., Kononen, J., Bubendorf, L., Chen, Y., Bittner, M. L., Torhorst, J., Haas, P., Bucher, C., Sauter, G., Kallioniemi, O. P. and Kallioniemi, A. (2000). Detecting activation of ribosomal protein S6 kinase by complementary DNA and tissue microarray analysis. Journal of the National Cancer Institute. 92: 1252-1259.

~~Barnes, C. J. and Kumar, R. (2003). Epidermal growth factor receptor family tyrosine kinases as signal integrators and therapeutic targets. Cancer & Metastasis Reviews 22: 301-307.~~

~~Barnes, C. J. and Kumar, R. (2004). Biology of the epidermal growth factor receptor family. Cancer Treatment & Research 119: 1-13.~~

~~Barr, F. G. (1998). Translocations, cancer and the puzzle of specificity. Nature Genetics 19: 121-124.~~

Barrett, T., Troup, D. B., Wilhite, S. E., Ledoux, P., Rudnev, D., Evangelista, C., Kim, I. F., Soboleva, A., Tomashevsky, M. and Edgar, R. (2007). NCBI GEO: mining tens of millions of expression profiles--database and tools update. Nucleic Acids Research 35: D760-765.

~~Barth, T., Moller, P. and Mechtersheimer, G. (1995). Differential expression of beta 1, beta 3 and beta 4 integrins in sarcomas of the small, round, blue cell category. Virchows Archiv 426: 19-25.~~

Baselga, J., Albanell, J., Ruiz, A., Lluch, A., Gascon, P., Guillem, V., Gonzalez, S., Sauleda, S., Marimon, I., Tabernero, J. M., Koehler, M. T. and Rojo, F. (2005). Phase II and tumor pharmacodynamic study of gefitinib in patients with advanced breast cancer. Journal of Clinical Oncology 23: 5323-5333.

~~Baselga, J. and Arteaga, C. L. (2005). Critical Update and Emerging Trends in Epidermal Growth Factor Receptor Targeting in Cancer. Journal of Clinical Oncology 23: 2445-2459.~~

~~Bates, S. E. and Fojo, T. (2005). Epidermal Growth Factor Receptor Inhibitors: A Moving Target? Clinical Cancer Research 11: 7203-7205.~~

Bayani, J., Zielenska, M., Pandita, A., Al-Romaih, K., Karaskova, J., Harrison, K., Bridge, J. A., Sorensen, P., Thorner, P. and Squire, J. A. (2003). Spectral karyotyping identifies recurrent complex rearrangements of chromosomes 8, 17, and 20 in osteosarcomas. Genes, Chromosomes & Cancer 36: 7-16.

~~417~~

~~Bibliography~~

Beech, D., Pollock, R. E., Tsan, R. and Radinsky, R. (1998). Epidermal growth factor receptor and insulin-like growth factor-I receptor expression and function in human soft-tissue sarcoma cells. International Journal of Oncology 12: 329-336.

Beer, D. G., Kardia, S. L., Huang, C. C., Giordano, T. J., Levin, A. M., Misek, D. E., Lin, L., Chen, G., Gharib, T. G., Thomas, D. G., Lizyness, M. L., Kuick, R., Hayasaka, S., Taylor, J. M., Iannettoni, M. D., Orringer, M. B. and Hanash, S. (2002). Gene- expression profiles predict survival of patients with lung adenocarcinoma. Nature Medicine 8: 816-824.

Beghini, A., Tibiletti, M. G., Roversi, G., Chiaravalli, A. M., Serio, G., Capella, C. and Larizza, L. (2001). Germline mutation in the juxtamembrane domain of the kit gene in a family with gastrointestinal stromal tumors and urticaria pigmentosa. Cancer 92: 657- 662.

Bell, D. W., Varley, J. M., Szydlo, T. E., Kang, D. H., Wahrer, D. C., Shannon, K. E., Lubratovich, M., Verselis, S. J., Isselbacher, K. J., Fraumeni, J. F., Birch, J. M., Li, F. P., Garber, J. E. and Haber, D. A. (1999). Heterozygous germ line hCHK2 mutations in Li-Fraumeni syndrome. Science 286: 2528-2531.

Benassi, M. S., Gamberi, G., Magagnoli, G., Molendini, L., Ragazzini, P., Merli, M., Chiesa, F., Balladelli, A., Manfrini, M., Bertoni, F., Mercuri, M. and Picci, P. (2001). Metalloproteinase expression and prognosis in soft tissue sarcomas. Annals of Oncology. 12: 75-80.

Benassi, M. S., Ragazzini, P., Gamberi, G., Sollazzo, M. R., Molendini, L., Ferrari, C., Merli, M., Bohling, T. and Picci, P. (1998). Adhesion molecules in high-grade soft tissue sarcomas: correlation to clinical outcome. European Journal of Cancer 34: 496- 502.

~~Benjamini, Y. and Hochberg, Y. (1995). Controlling the False Discovery Rate: a Practical and Powerful Approach to Multiple Testing. Journal of the Royal Statistical Society 57: 289-300.~~

Berardi, A. C., Parafioriti, A., Barisani, D., Papp, B., Armiraglio, E., Martinoli, M., Dalpra, L. and Santoro, A. (2003). A new human cell line, PDSS-26, from poorly differentiated synovial sarcoma, with unique chromosomal anomalies. Cancer Genetics & Cytogenetics 146: 116-124.

~~Bernards, R. (2003). Cues for Migration. Nature 425: 247-248.~~

Bernstein, C., Bernstein, H., Payne, C. M. and Garewal, H. (2002). DNA repair/pro- apoptotic dual-role proteins in five major DNA repair pathways: fail-safe protection against carcinogenesis. Mutation Research 511: 145-178.

Beroud, C., Collod-Beroud, G., Boileau, C., Soussi, T. and Junien, C. (2000). UMD (Universal mutation database): a generic software to build and analyze locus-specific databases. Human Mutation 15: 86-94.

~~418~~

~~Bibliography~~

~~Beroud, C. and Soussi, T. (1998). p53 gene mutation: software and database. Nucleic Acids Research 26: 200-204.~~

~~Beroud, C. and Soussi, T. (2003). The UMD-p53 database: new mutations and analysis tools. Human Mutation 21: 176-181.~~

Bertucci, F., Nasser, V., Granjeaud, S., Eisinger, F., Adelaide, J., Tagett, R., Loriod, B., Giaconia, A., Benziane, A., Devilard, E., Jacquemier, J., Viens, P., Nguyen, C., Birnbaum, D. and Houlgatte, R. (2002). Gene expression profiles of poor-prognosis primary breast cancer correlate with survival. Human Molecular Genetics. 11: 863-872.

~~Bertucci, F., Viens, P., Tagett, R., Nguyen, C., Houlgatte, R. and Birnbaum, D. (2003). DNA arrays in clinical oncology: promises and challenges. Laboratory Investigation 83: 305-316.~~

Bhattacharjee, A., Richards, W. G., Staunton, J., Li, C., Monti, S., Vasa, P., Ladd, C., Beheshti, J., Bueno, R., Gillette, M., Loda, M., Weber, G., Mark, E. J., Lander, E. S., Wong, W., Johnson, B. E., Golub, T. R., Sugarbaker, D. J. and Meyerson, M. (2001). Classification of human lung carcinomas by mRNA expression profiling reveals distinct adenocarcinoma subclasses. Proceedings of the National Academy of Sciences of the United States of America 98: 13790-13795.

~~Bianco, R., Melisi, D., Ciardiello, F. and Tortora, G. (2006). Key cancer cell signal transduction pathways as therapeutic targets. European Journal of Cancer 42: 290-294.~~

Bieche, I., Lerebours, F., Tozlu, S., Espie, M., Marty, M. and Lidereau, R. (2004). Molecular Profiling of Inflammatory Breast Cancer: Identification of a Poor-Prognosis Gene Expression Signature. Clinical Cancer Research 10: 6789-6795.

Birch, J. M., Hartley, A. L., Tricker, K. J., Prosser, J., Condie, A., Kelsey, A. M., Harris, M., Morris Jones, P. H., Binchy, A., Crowther, D., Craft, A. W., Eden, O. B., Evans, D. G. R., Thompson, E., Mann, J. R., Martin, J., Mitchell, E. L. D. and Santibanez-Koref, M. F. (1994). Prevalence and diversity of constitutional mutations in the p53 gene among 21 Li-Fraumeni families. Cancer Research 54: 1298-1304.

Blay, J. Y., Ray-Coquard, I., Alberti, L. and Ranchere, D. (2004). Targeting other abnormal signaling pathways in sarcoma: EGFR in synovial sarcomas, PPAR-gamma in liposarcomas. Cancer Treatment & Research 120: 151-167.

Blay, P., Astudillo, A., Buesa, J. M., Campo, E., Abad, M., Garcia-Garcia, J., Miquel, R., Marco, V., Sierra, M., Losa, R., Lacave, A., Brana, A., Balbin, M. and Freije, J. M. P. (2004). Protein Kinase C theta Is Highly Expressed in Gastrointestinal Stromal Tumors But Not in Other Mesenchymal Neoplasias. Clinical Cancer Research 10: 4089-4095.

~~Blobel, C. P. (2000). Remarkable roles of proteolysis on and beyond the cell surface. Current Opinion Cell Biology 12: 606-612.~~

~~419~~

~~Bibliography~~

Bolstad, B. M., Irizarry, R. A., Astrand, M. and Speed, T. P. (2003). A comparison of normalization methods for high density oligonucleotide array data based on variance and bias. Bioinformatics 19: 185-193.

Borden, E. C., Baker, L. H., Bell, R. S., Bramwell, V., Demetri, G. D., Eisenberg, B. L., Fletcher, C. D., Fletcher, J. A., Ladanyi, M., Meltzer, P., O'Sullivan, B., Parkinson, D. R., Pisters, P. W., Saxman, S., Singer, S., Sundaram, M., van Oosterom, A. T., Verweij, J., Waalen, J., Weiss, S. W. and Brennan, M. F. (2003). Soft tissue sarcomas of adults: state of the translational science. Clinical Cancer Research. 9: 1941-1956.

Bougeard, G., Limacher, J. M., Martin, C., Charbonnier, F., Killian, A., Delattre, O., Longy, M., Jonveaux, P., Fricker, J. P., Stoppa-Lyonnet, D., Flaman, J. M. and Frebourg, T. (2001). Detection of 11 germline inactivating TP53 mutations and absence of TP63 and HCHK2 mutations in 17 French families with Li-Fraumeni or Li- Fraumeni-like syndrome. Journal of Medical Genetics 38: 253-7.

~~Bowman, T., Garcia, R., Turkson, J. and Jove, R. (2000). STATs in oncogenesis. Oncogene 19: 2474-2488.~~

Brazma, A., Hingamp, P., Quackenbush, J., Sherlock, G., Spellman, P., Stoeckert, C., Aach, J., Ansorge, W., Ball, C. A., Causton, H. C., Gaasterland, T., Glenisson, P., Holstege, F. C., Kim, I. F., Markowitz, V., Matese, J. C., Parkinson, H., Robinson, A., Sarkans, U., Schulze-Kremer, S., Stewart, J., Taylor, R., Vilo, J. and Vingron, M. (2001). Minimum information about a microarray experiment (MIAME)-toward standards for microarray data. Nature Genetics. 29: 365-371.

Brezniceanu, M. L., Volp, K., Bosser, S., Solbach, C., Lichter, P., Joos, S. and Zornig, M. (2003). HMGB1 inhibits cell death in yeast and mammalian cells and is abundantly expressed in human breast carcinoma. FASEB Journal 17: 1295-1297.

~~Bromberg, J. and Darnell, J. E., Jr. (2000). The role of STATs in transcriptional control and their impact on cellular function. Oncogene 19: 2468-2473.~~

~~Brooks, J. J. (1986). The significance of double phenotypic patterns and markers in human sarcomas. A new model of mesenchymal differentiation. American Journal of Pathology 125: 113-123.~~

Brown, C. W., Houston-Hawkins, D. E., Woodruff, T. K. and Matzuk, M. M. (2000). Insertion of Inhbb into the Inhba locus rescues the Inhba-null phenotype and reveals new activin functions. Nature Genetics 25: 453-457.

~~Brown, H. (2003). The real value of microarray technology. Lancet Oncology. 4: 326.~~

~~Brownie, J., Shawcross, S., Theaker, J., Whitcombe, D., Ferrie, R., Newton, C. and Little, S. (1997). The elimination of primer-dimer accumulation in PCR. Nucleic Acids Research. 25: 3235-3241.~~

~~420~~

~~Bibliography~~

Bryan, T. M., Englezou, A., Dalla-Pozza, L., Dunham, M. A. and Reddel, R. R. (1997). Evidence for an alternative mechanism for maintaining telomere length in human tumors and tumor-derived cell lines. Nature Medicine 3: 1271-1274.

~~Bryan, T. M., Englezou, A., Gupta, J., Bacchetti, S. and Reddel, R. R. (1995). Telomere elongation in immortal human cells without detectable telomerase activity. EMBO Journal 14: 4240-4248.~~

Bubendorf, L., Kononen, J., Koivisto, P., Schraml, P., Moch, H., Gasser, T. C., Willi, N., Mihatsch, M. J., Sauter, G. and Kallioniemi, O. P. (1999). Survey of gene amplifications during prostate cancer progression by high-throughout fluorescence in situ hybridization on tissue microarrays. Cancer Research 59: 803-806.

~~Burger, A. M. (2007). Highlights in experimental therapeutics. Cancer Letters 245: 11- 21.~~

Burger, A. M., Dai, F., Schultes, C. M., Reszka, A. P., Moore, M. J., Double, J. A. and Neidle, S. (2005). The G-quadruplex-interactive molecule BRACO-19 inhibits tumor growth, consistent with telomere targeting and interference with telomerase function. Cancer Research 65: 1489-1496.

Burt, E. C., McGown, G., Thorncroft, M., James, L. A., Birch, J. M. and Varley, J. M. (1999). Exclusion of the genes CDKN2 and PTEN as causative gene defects in Li- Fraumeni syndrome. British Journal of Cancer 80: 9-10.

~~Bustin, S. A. (2000). Absolute quantification of mRNA using real-time reverse transcription polymerase chain reaction assays. Journal of Molecular Endocrinology 25: 169-93.~~

~~Bustin, S. A. (2002). Quantification of mRNA using real-time reverse transcription PCR (RT-PCR): trends and problems. Journal of Molecular Endocrinology 29: 23-39.~~

Busund, L. T., Ow, K. T., Russell, P., Crowe, P. J. and Yang, J. L. (2004). Expression of insulin-like growth factor mitogenic signals in adult soft-tissue sarcomas: significant correlation with malignant potential. Virchows Archives 444: 142-148.

Caja, L., Ortiz, C., Bertran, E., Murillo, M. M., Miro-Obradors, M. J., Palacios, E. and Fabregat, I. (2007). Differential intracellular signalling induced by TGF-[beta] in rat adult hepatocytes and hepatoma cells: Implications in liver carcinogenesis. Cellular Signalling 19: 683-694.

Calo, V., Migliavacca, M., Bazan, V., Macaluso, M., Buscemi, M., Gebbia, N. and Russo, A. (2003). STAT proteins: from normal control of cellular events to tumorigenesis. Journal of Cell Physiology 197: 157-168.

~~Camp, R. L., Charette, L. A. and Rimm, D. L. (2000). Validation of Tissue Microarray Technology in Breast Carcinoma. Laboratory Investigation 80: 1943-1949.~~

~~421~~

~~Bibliography~~

~~Camp, R. L., Chung, G. G. and Rimm, D. L. (2002). Automated subcellular localization and quantification of protein expression in tissue microarrays. Nature Medicine 8: 1323- 1327.~~

Cance, W. G., Brennan, M. F., Dudas, M. E., Huang, C. M. and Cordon-Cardo, C. (1990). Altered expression of the retinoblastoma gene product in human sarcomas. New England Journal of Medicine 323: 1457-1462.

~~Cangul, H. (2004). Hypoxia upregulates the expression of the NDRG1 gene leading to its overexpression in various human cancers. BMC Genetics 5: 2.~~

Cano, A., Perez-Moreno, M. A., Rodrigo, I., Locascio, A., Blanco, M. J., del Barrio, M. G., Portillo, F. and Nieto, M. A. (2000). The transcription factor snail controls epithelial-mesenchymal transitions by repressing E-cadherin expression. Nature Cell Biology 2: 76-83.

Cardone, M. H., Roy, N., Stennicke, H. R., Salvesen, G. S., Franke, T. F., Stanbridge, E., Frisch, S. and Reed, J. C. (1998). Regulation of Cell Death Protease Caspase-9 by Phosphorylation. Science 282: 1318-1321.

~~Carmeliet, P. and Jain, R. K. (2000). Angiogenesis in cancer and other diseases. Nature 407: 249-257.~~

Carroll, P. E., Okuda, M., Horn, H. F., Biddinger, P., Stambrook, P. J., Gleich, L. L., Li, Y. Q., Tarapore, P. and Fukasawa, K. (1999). Centrosome hyperamplification in human cancer: chromosome instability induced by p53 mutation and/or Mdm2 overexpression. Oncogene 18: 1935-1944.

Catlett-Falcone, R., Landowski, T. H., Oshiro, M. M., Turkson, J., Levitzki, A., Savino, R., Ciliberto, G., Moscinski, L., Fernandez-Luna, J. L., Nunez, G., Dalton, W. S. and Jove, R. (1999). Constitutive activation of Stat3 signaling confers resistance to apoptosis in human U266 myeloma cells. Immunity 10: 105-115.

~~Chan, J. K. C., Fletcher, C. D. M., Nayler, S. J. and Cooper, K. (1997). Follicular dendritic cell sarcoma. Cancer 79: 294-313.~~

Chan, P. M., Ilangumaran, S., La Rose, J., Chakrabartty, A. and Rottapel, R. (2003). Autoinhibition of the Kit Receptor Tyrosine Kinase by the Cytosolic Juxtamembrane Region. Molecular and Cellular Biology 23: 3067-3078.

Chang, J. T., Wang, H. M., Chang, K. W., Chen, W. H., Wen, M. C., Hsu, Y. M., Yung, B. Y., Chen, I. H., Liao, C. T., Hsieh, L. L. and Cheng, A. J. (2005). Identification of differentially expressed genes in oral squamous cell carcinoma (OSCC): overexpression of NPM, CDK1 and NDRG1 and underexpression of CHES1. International Journal of Cancer 114: 942-949.

Chao, C., Al-Saleem, T., Brooks, J. J., Rogatko, A. and Kraybill, W. G. E., B. (2001). Vascular endothelial growth factor and soft tissue sarcomas: tumor expression correlates with grade. Annals of Surgical Oncology. 8: 260-7.

~~422~~

~~Bibliography~~

~~Chen, Y.-G., Lui, H. M., Lin, S.-L., Lee, J. M. and Ying, S.-Y. (2002). Regulation of Cell Proliferation, Apoptosis, and Carcinogenesis by Activin. Experimental Biology and Medicine 227: 75-87.~~

~~Cheung, I. Y. and Cheung, N.-K. V. (2001). Quantitation of Marrow Disease in Neuroblastoma by Real-Time Reverse Transcription-PCR. Clinical Cancer Research 7: 1698-1705.~~

~~Cheung, V. G. and Spielman, R. S. (2002). The genetics of variation in gene expression. Nature Genetics. 32: 522-525.~~

Choi, Y. R., Kim, H., Kang, H. J., Kim, N. G., Kim, J. J., Park, K. S., Paik, Y. K. and Kim, H. O. (2003). Overexpression of high mobility group box 1 in gastrointestinal stromal tumors with KIT mutation. Cancer Research 63: 2188-2193.

Chuaqui, R. F., Bonner, R. F., Best, C. J., Gillespie, J. W., Flaig, M. J., Hewitt, S. M., Phillips, J. L., Krizman, D. B., Tangrea, M. A., Ahram, M., Linehan, W. M., Knezevic, V. and Emmert-Buck, M. R. (2002). Post-analysis follow-up and validation of microarray experiments. Nature Genetics 32: 509-514.

~~Churchill, G. A. (2002). Fundamentals of experimental design for cDNA microarrays. Nature Genetics. 32: 490-495.~~

~~Ciardiello, F., De Vita, F., Orditura, M. and Tortora, G. (2004). The role of EGFR inhibitors in nonsmall cell lung cancer. Current Opinion in Oncology 16: 130-135.~~

~~Ciardiello, F. and Tortora, G. (2001). A Novel Approach in the Treatment of Cancer: Targeting the Epidermal Growth Factor Receptor. Clinical Cancer Research 7: 2958- 2970.~~

Cohen, C. D., Frach, K., Schlöndorff, D. and Kretzler, M. (2002). Quantitative gene expression analysis in renal biopsies: A novel protocol for a high-throughput multicenter application. Kidney International 61: 133-140.

Coindre, J., Terrier, P., Bui, N., Bonichon, F., Collin, F., Le Doussal, V., Mandard, A., Vilain, M., Jacquemier, J., Duplay, H., Sastre, X., Barlier, C., Henry-Amar, M., Mace- Lesech, J. and Contesso, G. (1996). Prognostic factors in adult patients with locally controlled soft tissue sarcoma. A study of 546 patients from the French Federation of Cancer Centers Sarcoma Group. Journal of Clinical Oncology 14: 869-877.

~~Coindre, J. M. (1993). Pathology and grading of soft tissue sarcomas. Cancer Treatment & Research 67: 1-22.~~

~~Coindre, J. M., Nguyen, B. B., Bonichon, F., de Mascarel, I. and Trojani, M. (1988). Histopathologic grading in spindle cell soft tissue sarcomas. Cancer 61: 2305-2309.~~

~~Coindre, J. M., Terrier, P., Guillou, L., Le Doussal, V., Collin, F., Ranchere, D., Sastre, X., Vilain, M. O., Bonichon, F. and N'Guyen Bui, B. (2001). Predictive value of grade~~

~~423~~

~~Bibliography~~

~~for metastasis development in the main histologic types of adult soft tissue sarcomas: a study of 1240 patients from the French Federation of Cancer Centers Sarcoma Group. Cancer 91: 1914-1926.~~

Coindre, J. M., Trojani, M., Contesso, G., David, M., Rouesse, J., Bui, N. B., Bodaert, A., De Mascarel, I., De Mascarel, A. and Goussot, J. F. (1986). Reproducibility of a histopathologic grading system for adult soft tissue sarcoma. Cancer 58: 306-309.

~~Colgin, L. M. and Reddel, R. R. (1999). Telomere maintenance mechanisms and cellular immortalization. Current Opinion in Genetics & Development 9: 97-103.~~

Colgin, L. M., Wilkinson, C., Englezou, A., Kilian, A., Robinson, M. O. and Reddel, R. R. (2000). The hTERTalpha splice variant is a dominant negative inhibitor of telomerase activity. Neoplasia 2: 426-32.

~~Corless, C. L., Fletcher, J. A. and Heinrich, M. C. (2004). Biology of gastrointestinal stromal tumors. Journal of Clinical Oncology 22: 3813-3825.~~

~~Cormier, J. N. and Pollock, R. E. (2004). Soft tissue sarcomas. Ca: a Cancer Journal for Clinicians. 54: 94-109.~~

Costa, A., Daidone, M. G., Daprai, L., Villa, R., Cantu, S., Pilotti, S., Mariani, L., Gronchi, A., Henson, J. D., Reddel, R. R. and Zaffaroni, N. (2006). Telomere Maintenance Mechanisms in Liposarcomas: Association with Histologic Subtypes and Disease Progression. Cancer Research 66: 8918-8924.

~~Costa, J., Wesley, R. A., Glatstein, E. and Rosenberg, S. A. (1984). The grading of soft tissue sarcomas. Results of a clinicohistopathologic correlation in a series of 163 cases. Cancer 53: 530-541.~~

~~Cotran, R. S., Kumar, V. and Robbins, S. L. (1994). Neoplasia: Mechanisms of invasion and metastases. Robbins Pathologic Basis of Disease. F. J. Schoen. Philadelphia, W.B. Saunders Company: 276-280.~~

~~Crawford, A. H. (1986). Neurofibromatosis in children. Acta Orthopaedica Scandinavica. Supplementum. 57: 9-60.~~

~~Creager, A. J., Cohen, J. A. and Geradts, J. (2001). Aberrant expression of cell-cycle regulatory proteins in human mesenchymal neoplasia. Cancer Detection & Prevention 25: 123-131.~~

Crnalic, S., Panagopoulos, I., Boquist, L., Mandahl, N., Stenling, R. and Lofvenberg, R. (2002). Establishment and characterisation of a human clear cell sarcoma model in nude mice. International Journal of Cancer. 101: 505-511.

Crnogorac-Jurcevic, T., Efthimiou, E., Nielsen, T., Loader, J., Terris, B., Stamp, G., Baron, A., Scarpa, A. and Lemoine, N. R. (2002). Expression profiling of microdissected pancreatic adenocarcinomas. Oncogene 21: 4587-4594.

~~424~~

~~Bibliography~~

~~Crowe, F. W., Schull, W. J. and Neel, J. V. (1956). A Clinical, Pathological and Genetic Study of Multiple Neurofibromatosis. Springfield, Ill, Charles C Thomas.~~

D'Adamo, D. R., Anderson, S. E., Albritton, K., Yamada, J., Riedel, E., Scheu, K., Schwartz, G. K., Chen, H. and Maki, R. G. (2005). Phase II Study of Doxorubicin and Bevacizumab for Patients With Metastatic Soft-Tissue Sarcomas. Journal of Clinical Oncology 23: 7135-7142.

D'Agostino, A. N., Soule, E. H. and Miller, R. H. (1963). Sarcomas of the peripheral nerves and somatic soft tissues associated with multiple neurofibromatosis (Von Recklinghausen's disease). Cancer 16: 1015-1027.

D'Souza, R. J., Sheikh, Z. A., Busund, L. T., Russell, P. J., Crowe, P. J. and Yang, J. L. (2003). Expression of nm23 protein in adult soft tissue sarcoma is correlated with histological grade. Anticancer Research 23: 3289-3294.

Daneshmand, M., Parolin, D. A. E., Hirte, H. W., Major, P., Goss, G., Stewart, D., Batist, G., Miller, W. H., Jr., Matthews, S., Seymour, L. and Lorimer, I. A. J. (2003). A Pharmacodynamic Study of the Epidermal Growth Factor Receptor Tyrosine Kinase Inhibitor ZD1839 in Metastatic Colorectal Cancer Patients. Clinical Cancer Research 9: 2457-2464.

~~Darnell, J., Jr, Kerr, I. and Stark, G. (1994). Jak-STAT pathways and transcriptional activation in response to IFNs and other extracellular signaling proteins. Science 264: 1415-1421.~~

~~Daub, H., Weiss, F. U., Wallasch, C. and Ullrich, A. (1996). Role of transactivation of the EGF receptor in signalling by G-protein-coupled receptors. Nature 379: 557-560.~~

Daw, N. C., Furman, W. L., Stewart, C. F., Iacono, L. C., Krailo, M., Bernstein, M. L., Dancey, J. E., Speights, R. A., Blaney, S. M., Croop, J. M., Reaman, G. H. and Adamson, P. C. (2005). Phase I and Pharmacokinetic Study of Gefitinib in Children With Refractory Solid Tumors: A Children's Oncology Group Study. Journal of Clinical Oncology 23: 6172-6180.

Dawson, N. A., Guo, C., Zak, R., Dorsey, B., Smoot, J., Wong, J. and Hussain, A. (2004). A phase II trial of gefitinib (Iressa, ZD1839) in stage IV and recurrent renal cell carcinoma. Clinical Cancer Research 10: 7812-7819. de Alava, E., Antonescu, C. R., Panizo, A., Leung, D., Meyers, P. A., Huvos, A. G., Pardo-Mindan, F. J., Healey, J. H. and Ladanyi, M. (2000). Prognostic impact of P53 status in Ewing sarcoma. Cancer 89: 783-792.

~~De Giorgi, U. and Verweij, J. (2005). Imatinib and gastrointestinal stromal tumors: Where do we go from here? Molecular Cancer Therapeutics 4: 495-501.~~

De Luca, A., Carotenuto, A., Rachiglio, A., Gallo, M., Maiello, M. R., Aldinucci, D., Pinto, A., Normanno, N., De Luca, A., Carotenuto, A., Rachiglio, A., Gallo, M., Maiello, M. R., Aldinucci, D., Pinto, A. and Normanno, N. (2008). The role of the

~~425~~

~~Bibliography~~

~~EGFR signaling in tumor microenvironment. Journal of Cellular Physiology 214: 559- 67.~~

Debiec-Rychter, M., Cools, J., Dumez, H., Sciot, R., Stul, M., Mentens, N., Vranckx, H., Wasag, B., Prenen, H., Roesel, J., Hagemeijer, A., Van Oosterom, A. and Marynen, P. (2005). Mechanisms of resistance to imatinib mesylate in gastrointestinal stromal tumors and activity of the PKC412 inhibitor against imatinib-resistant mutants. Gastroenterology 128: 270-279.

Debiec-Rychter, M., Dumez, H., Judson, I., Wasag, B., Verweij, J., Brown, M., Dimitrijevic, S., Sciot, R., Stul, M., Vranck, H., Scurr, M., Hagemeijer, A., van Glabbeke, M. and van Oosterom, A. T. (2004). Use of c-KIT/PDGFRA mutational analysis to predict the clinical response to imatinib in patients with advanced gastrointestinal stromal tumours entered on phase I and II studies of the EORTC Soft Tissue and Bone Sarcoma Group. European Journal of Cancer 40: 689-695.

Debiec-Rychter, M., Sciot, R., Le Cesne, A., Schlemmer, M., Hohenberger, P., van Oosterom, A. T., Blay, J. Y., Leyvraz, S., Stul, M., Casali, P. G., Zalcberg, J., Verweij, J., Van Glabbeke, M., Hagemeijer, A., Judson, I., Tissue, E. S., Bone Sarcoma, G., The Italian Sarcoma, G. and Australasian GastroIntestinal Trials, G. (2006). KIT mutations and dose selection for imatinib in patients with advanced gastrointestinal stromal tumours. European Journal of Cancer 42: 1093-1103.

Debrock, G., Vanhentenrijk, V., Sciot, R., Debiec-Rychter, M., Oyen, R. and Van Oosterom, A. (2003). A phase II trial with rosiglitazone in liposarcoma patients. British Journal of Cancer 89: 1409-1412.

DeMatteo, R. P. (2002). The GIST of Targeted Cancer Therapy: A Tumor (Gastrointestinal Stromal Tumor), a Mutated Gene (c-kit), and a Molecular Inhibitor (STI571). Annals of Surgical Oncology. 9: 831-839.

~~DeMatteo, R. P., Heinrich, M. C., El-Rifai, W. e. M. and Demetri, G. (2002). Clinical management of gastrointestinal stromal tumors: Before and after STI-571. Human Pathology 33: 466-477.~~

DeMatteo, R. P., Lewis, J. J., Leung, D., Mudan, S. S., Woodruff, J. M. and Brennan, M. F. (2000). Two Hundred Gastrointestinal Stromal Tumors: Recurrence Patterns and Prognostic Factors for Survival. Annals of Surgery. 231: 51-58.

Demetri, G. D. (2002). Targeting the molecular pathophysiology of gastrointestinal stromal tumors with imatinib. Mechanisms, successes, and challenges to rational drug development. Hematol Oncol Clin North Am 16: 1115-1124.

Demetri, G. D., Fletcher, C. D. M., Mueller, E., Sarraf, P., Naujoks, R., Campbell, N., Spiegelman, B. M. and Singer, S. (1999). Induction of solid tumor differentiation by the peroxisome proliferator-activated receptor-gamma ligand troglitazone in patients with liposarcoma. Proceedings of the National Academy of Sciences of the United States of America 96: 3951-3956.

~~426~~

~~Bibliography~~

Demetri, G. D., van Oosterom, A. T., Garrett, C. R., Blackstein, M. E., Shah, M. H., Verweij, J., McArthur, G., Judson, I. R., Heinrich, M. C., Morgan, J. A., Desai, J., Fletcher, C. D., George, S., Bello, C. L., Huang, X., Baum, C. M. and Casali, P. G. (2006). Efficacy and safety of sunitinib in patients with advanced gastrointestinal stromal tumour after failure of imatinib: a randomised controlled trial. Lancet 368: 1329-1338.

Dempsey, C. E., Sakurai, H., Sugita, T. and Guesdon, F. (2000). Alternative splicing and gene structure of the transforming growth factor beta-activated kinase 1. Biochimica et Biophysica Acta 1517: 46-52.

Derre, J., Lagace, R., Nicolas, A., Mairal, A., Chibon, F., Coindre, J. M., Terrier, P., Sastre, X. and Aurias, A. (2001). Leiomyosarcomas and most malignant fibrous histiocytomas share very similar comparative genomic hybridization imbalances: an analysis of a series of 27 leiomyosarcomas. Laboratory Investigation 81: 211-215.

~~Derynck, R. and Zhang, Y. E. (2003). Smad-dependent and Smad-independent pathways in TGF-[beta] family signalling. Nature 425: 577-584.~~

Desai, J., Maki, R., Heinrich, M. C., Fletcher, J. A., Ryan, D., Morgan, J. A., St. Amand, M., George, S., Manola, J., Baum, C., Cohen, D. P., Corless, C. L., Fletcher, C. D., Van den Abbeele, A. and Demetri, G. D. (2004). Activity and tolerability of the multi-targeted tyrosine kinase inhibitor SU011248 in patients (pts) with metastatic gastrointestinal stromal tumor (GIST) refractory to imatinib mesylate. ASCO 2004. 2004 Gastrointestinal Cancers Symposium.

~~Dhaene, K., Van Marck, E. and Parwaresch, R. (2000). Telomeres, telomerase and cancer: an up-date. Virchows Archives 437: 1-16.~~

Dhanasekaran, S. M., Barrette, T. R., Ghosh, D., Shah, R., Varambally, S., Kurachi, K., Pienta, K. J., Rubin, M. A. and Chinnaiyan, A. M. (2001). Delineation of prognostic biomarkers in prostate cancer. Nature 412: 822-826.

~~Di Persio, J., Brennan, J., Lichtman, M. and Speiser, B. (1978). Human cell lines that elaborate colony-stimulating activity for the marrow cells of man and other species. Blood 51: 507-19.~~

~~Dibb, N. J., Dilworth, S. M. and Mol, C. D. (2004). Switching on kinases: oncogenic activation of BRAF and the PDGFR family. Nature Reviews Cancer 4: 718-727.~~

~~Dobbin, K., Shih, J. H. and Simon, R. (2003). Statistical design of reverse dye microarrays. Bioinformatics. 19: 803-810.~~

~~Dobbin, K. and Simon, R. (2002). Comparison of microarray designs for class comparison and class discovery. Bioinformatics. 18: 1438-1445.~~

Dolled-Filhart, M., Camp, R. L., Kowalski, D. P., Smith, B. L. and Rimm, D. L. (2003). Tissue Microarray Analysis of Signal Transducers and Activators of Transcription 3 (Stat3) and Phospho-Stat3 (Tyr705) in Node-negative Breast Cancer Shows Nuclear

~~427~~

~~Bibliography~~

~~Localization Is Associated with a Better Prognosis. Clinical Cancer Research 9: 594- 600.~~

Dong, J. T., Lamb, P. W., Rinker-Schaeffer, C. W., Vukanovic, J., Ichikawa, T., Isaacs, J. T. and Barrett, J. C. (1995). KAI1, a metastasis suppressor gene for prostate cancer on human chromosome 11p11.2. Science 268: 884-886.

Donninger, H., Bonome, T., Radonovich, M., Pise-Masison, C. A., Brady, J., Shih, J. H., Barrett, J. C. and Birrer, M. J. (2004). Whole genome expression profiling of advance stage papillary serous ovarian cancer reveals activated pathways. Oncogene 23: 8065-8077.

Drube, S., Stirnweiss, J., Valkova, C. and Liebmann, C. (2006). Ligand-independent and EGF receptor-supported transactivation: lessons from beta2-adrenergic receptor signalling. Cellular Signalling 18: 1633-1646.

Duan, Z., Foster, R., Bell, D. A., Mahoney, J., Wolak, K., Vaidya, A., Hampel, C., Lee, H. and Seiden, M. V. (2006). Signal Transducers and Activators of Transcription 3 Pathway Activation in Drug-Resistant Ovarian Cancer. Clinical Cancer Research 12: 5055-5063.

Dudoit, S., Yang, Y. H., Callow, M. J. and Speed, T. (2002). Statistical methods for identifying differentially expressed genes in replicated cDNA microarray experiments. Statistical Sinica 12: 111-139.

~~Duensing, A., Heinrich, M. C., Fletcher, C. D. and Fletcher, J. A. (2004). Biology of gastrointestinal stromal tumors: KIT mutations and beyond. Cancer Investigation 22: 106-116.~~

Duensing, A., Joseph, N. E., Medeiros, F., Smith, F., Hornick, J. L., Heinrich, M. C., Corless, C. L., Demetri, G. D., Fletcher, C. D. M. and Fletcher, J. A. (2004). Protein kinase C theta (PKC theta) expression and constitutive activation in gastrointestinal stromal tumors (GISTs). Cancer Research 64: 5127-5131.

Duensing, A., Medeiros, F., McConarty, B., Joseph, N. E., Panigrahy, D., Singer, S., Fletcher, C. D., Demetri, G. D. and Fletcher, J. A. (2004). Mechanisms of oncogenic KIT signal transduction in primary gastrointestinal stromal tumors (GISTs). Oncogene 23: 3999-4006.

Eberhard, D. A., Johnson, B. E., Amler, L. C., Goddard, A. D., Heldens, S. L., Herbst, R. S., Ince, W. L., Janne, P. A., Januario, T., Johnson, D. H., Klein, P., Miller, V. A., Ostland, M. A., Ramies, D. A., Sebisanovic, D., Stinson, J. A., Zhang, Y. R., Seshagiri, S. and Hillan, K. J. (2005). Mutations in the Epidermal Growth Factor Receptor and in KRAS Are Predictive and Prognostic Indicators in Patients With Non-Small-Cell Lung Cancer Treated With Chemotherapy Alone and in Combination With Erlotinib. Journal of Clinical Oncology 23: 5900-5909.

~~428~~

~~Bibliography~~

~~Edgar, R., Domrachev, M. and Lash, A. E. (2002). Gene Expression Omnibus: NCBI gene expression and hybridization array data repository. Nucleic Acids Research 30: 207-210.~~

~~Eilber, F. R., Morton, D. L., Eckardt, J., Grant, T. and Weisenburger, T. (1984). Limb salvage for skeletal and soft tissue sarcomas. Multidisciplinary preoperative therapy. Cancer 53: 2579-2584.~~

Eisen, M. B., Spellman, P. T., Brown, P. O. and Botstein, D. (1998). Cluster analysis and display of genome-wide expression patterns. Proceedings of the National Academy of Sciences of the United States of America 95: 14863-14868.

El-Rayes, B. F., Ali, S., Ali, I. F., Philip, P. A., Abbruzzese, J. and Sarkar, F. H. (2006). Potentiation of the Effect of Erlotinib by Genistein in Pancreatic Cancer: The Role of Akt and Nuclear Factor-{kappa}B. Cancer Research 66: 10553-10559.

El-Rifai, W. e., Sarlomo-Rikala, M., Andersson, L. C., Knuutila, S. and Miettinen, M. (2000). DNA Sequence Copy Number Changes in Gastrointestinal Stromal Tumors: Tumor Progression and Prognostic Significance. Cancer Research 60: 3899-3903.

~~Elliott, R. L. and Blobe, G. C. (2005). Role of Transforming Growth Factor Beta in Human Cancer. Journal of Clinical Oncology 23: 2078-2093.~~

Engellau, J., Akerman, M., Anderson, H., Domanski, H. A., Rambech, E., Alvegard, T. A. and Nilbert, M. (2001). Tissue microarray technique in soft tissue sarcoma: immunohistochemical Ki-67 expression in malignant fibrous histiocytoma. Applied Immunohistochemistry & Molecular Morphology. 9: 3583-3563.

Engellau, J., Bendahl, P. O., Persson, A., Domanski, H. A., Akerman, M., Gustafson, P., Alvegard, T. A., Nilbert, M. and Rydholm, A. (2005). Improved prognostication in soft tissue sarcoma: independent information from vascular invasion, necrosis, growth pattern, and immunostaining using whole-tumor sections and tissue microarrays. Human Pathology 36: 994-1002.

~~Enzinger, F. M. and Weiss, S. W. (1995a). General Considerations. Soft Tissue Tumors. F. M. Enzinger and S. W. Weiss. St Louis, Mosby: 1-16.~~

~~Enzinger, F. M. and Weiss, S. W. (1995b). Soft Tissue Tumours. St. Louis, Mosby.~~

Epstein, A., Martin, A. and Kempson, R. (1984). Use of a newly established human cell line (SU-CCS-1) to demonstrate the relationship of clear cell sarcoma to malignant melanoma. Cancer Research 44: 1265-1274.

~~Erlandson, R. A. and Antonescu, C. R. (2004). The rise and fall of malignant fibrous histiocytoma. Ultrastructural Pathology 28: 283-289.~~

~~Erlandsson Harris, H. and Andersson, U. (2004). Mini-review: The nuclear protein HMGB1 as a proinflammatory mediator. European Journal of Immunology. 34: 1503- 1512.~~

~~429~~

~~Bibliography~~

Eschrich, S., Yang, I., Bloom, G., Kwong, K. Y., Boulware, D., Cantor, A., Coppola, D., Kruhoffer, M., Aaltonen, L., Orntoft, T. F., Quackenbush, J. and Yeatman, T. J. (2005). Molecular Staging for Survival Prediction of Colorectal Cancer Patients. Journal of Clinical Oncology 23: 3526-3535.

Esni, F., Ghosh, B., Biankin, A. V., Lin, J. W., Albert, M. A., Yu, X., MacDonald, R. J., Civin, C. I., Real, F. X., Pack, M. A., Ball, D. W. and Leach, S. D. (2004). Notch inhibits Ptf1 function and acinar cell differentiation in developing mouse and zebrafish pancreas. Development 131: 4213-4224.

~~Evans, A., Lennard, T. W. and Davies, B. R. (2004). High-mobility group protein 1(Y): metastasis-associated or metastasis-inducing? Journal of Surgical Oncology 88: 86-99.~~

~~Evans, D. G., Baser, M. E., McGaughran, J., Sharif, S., Howard, E. and Moran, A. (2002). Malignant peripheral nerve sheath tumours in neurofibromatosis 1. Journal of Medical Genetics 39: 311-314.~~

~~Ewen, M. E. and Lamb, J. (2004). The activities of cyclin D1 that drive tumorigenesis. Trends in Molecular Medicine 10: 158-162.~~

Falcioni, R., Antonini, A., Nistico, P., Di Stefano, S., Crescenzi, M., Natali, P. G. and Sacchi, A. (1997). Alpha 6 beta 4 and alpha 6 beta 1 integrins associate with ErbB-2 in human carcinoma cell lines. Experimental Cell Research 236: 76-85.

~~Fischer, O. M., Hart, S., Gschwind, A. and Ullrich, A. (2003). EGFR signal transactivation in cancer cells. Biochemical Society Transactions 31: 1203-1208.~~

~~Fischer, O. M., Hart, S. and Ullrich, A. (2006). Dissecting the epidermal growth factor receptor signal transactivation pathway. Methods in Molecular Biology 327: 85-97.~~

~~Fleming, I. D. (2001). AJCC/TNM cancer staging, present and future. Journal of Surgical Oncology 77: 233-236.~~

Fleming, I. D., Phillips, J. L. and Menck, H. R. (1996). The National Cancer Data Base report on completeness of American Joint Committee on Cancer staging in United States cancer facilities. The American College of Surgeons Commission on Cancer and the American Cancer Society. Cancer 78: 1498-1504.

Fletcher, C. D., Dal Cin, P., de Wever, I., Mandahl, N., Mertens, F., Mitelman, F., Rosai, J., Rydholm, A., Sciot, R., Tallini, G., van den Berghe, H., Vanni, R. and Willen, H. (1999). Correlation between clinicopathological features and karyotype in spindle cell sarcomas. A report of 130 cases from the CHAMP study group. American Journal of Pathology 154: 1841-1847.

Fletcher, C. D., Gustafson, P., Rydholm, A., Willen, H. and Akerman, M. (2001). Clinicopathologic re-evaluation of 100 malignant fibrous histiocytomas: prognostic relevance of subclassification. Journal of Clinical Oncology. 19: 3045-3050.

~~430~~

~~Bibliography~~

Fletcher, C. D. M., Berman, J. J., Corless, C., Gorstein, F., Lasota, J., Longley, B. J., Miettinen, M., O'Leary, T. J., Remotti, H. and Rubin, B. P. (2002). Diagnosis of gastrointestinal stromal tumors: A consensus approach. Human Pathology 33: 459-465.

~~Fletcher, J. A. (2004). Molecular biology and cytogenetics of soft tissue sarcomas: relevance for targeted therapies. Cancer Treatment & Research 120: 99-116.~~

~~Fogh, J., Fogh, J. M. and Orfeo, T. (1977). One hundred and twenty-seven cultured human tumor cell lines producing tumours in nude mice. J Natl Cancer Inst 59: 221-6.~~

~~Folkman, J. (2002). Role of angiogenesis in tumor growth and metastasis. Seminars in Oncology 29: 15-8.~~

Francis, P., Namlos, H. M., Muller, C., Eden, P., Fernebro, J., Berner, J. M., Bjerkehagen, B., Akerman, M., Bendahl, P. O., Isinger, A., Rydholm, A., Myklebost, O. and Nilbert, M. (2007). Diagnostic and prognostic gene expression signatures in 177 soft tissue sarcomas: hypoxia-induced transcription profile signifies metastatic potential. BMC Genomics 8.

Frebourg, T., Barbier, N., Yan, Y. X., Garber, J. E., Dreyfus, M., Fraumeni, J., Jr., Li, F. P. and Friend, S. H. (1995). Germ-line p53 mutations in 15 families with Li-Fraumeni syndrome. American Journal of Human Genetics 56: 608-615.

Freier, K., Joos, S., Flechtenmacher, C., Devens, F., Benner, A., Bosch, F. X., Lichter, P. and Hofele, C. (2003). Tissue Microarray Analysis Reveals Site-specific Prevalence of Oncogene Amplifications in Head and Neck Squamous Cell Carcinoma. Cancer Research 63: 1179-1182.

~~Freshney, R. I. (2000). Intoduction to basic principles. Animal Cell Culture. J. W. Masters. Oxford, Oxford University Press: 1-15.~~

~~Friedl, P. and Wolf, K. (2003). Tumour-cell invasion and migration: diversity and escape mechanisms. Nature Reviews. Cancer. 3: 362-374.~~

Friend, S. H., Horowitz, J. M., Gerber, M. R., Wang, X. F., Bogenmann, E., Li, F. P. and Weinberg, R. A. (1987). Deletions of a DNA sequence in retinoblastomas and mesenchymal tumors: organization of the sequence and its encoded protein. Proceedings of the National Academy of Sciences of the United States of America 84: 9059-9063.

Fritz, B., Schubert, F., Wrobel, G., Schwaenen, C., Wessendorf, S., Nessling, M., Korz, C., Rieker, R. J., Montgomery, K., Kucherlapati, R., Mechtersheimer, G., Eils, R., Joos, S. and Lichter, P. (2002). Microarray-based copy number and expression profiling in dedifferentiated and pleomorphic liposarcoma. Cancer Research 62: 2993-2998.

Frolov, A., Chahwan, S., Ochs, M., Arnoletti, J. P., Pan, Z. Z., Favorova, O., Fletcher, J., von Mehren, M., Eisenberg, B. and Godwin, A. K. (2003). Response markers and the molecular mechanisms of action of Gleevec in gastrointestinal stromal tumors. Molecular Cancer Therapeutics 2: 699-709.

~~431~~

~~Bibliography~~

Frustaci, S., Gherlinzoni, F., De Paoli, A., Bonetti, M., Azzarelli, A., Comandone, A., Olmi, P., Buonadonna, A., Pignatti, G., Barbieri, E., Apice, G., Zmerly, H., Serraino, D. and Picci, P. (2001). Adjuvant Chemotherapy for Adult Soft Tissue Sarcomas of the Extremities and Girdles: Results of the Italian Randomized Cooperative Trial. Journal of Clinical Oncology 19: 1238-1247.

~~Fu, M., Wang, C., Li, Z., Sakamaki, T. and Pestell, R. G. (2004). Minireview: Cyclin D1: normal and abnormal functions. Endocrinology 145: 5439-5447.~~

~~Fukasawa, K., Wiener, F., Vande Woude, G. F. and Mai, S. (1997). Genomic instability and apoptosis are frequent in p53 deficient young mice. Oncogene 15: 1295-1302.~~

~~Fukuhara, S., Chikumi, H. and Gutkind, J. S. (2001). RGS-containing RhoGEFs: the missing link between transforming G proteins and Rho? Oncogene 20: 1661-1668.~~

Furey, T. S., Cristianini, N., Duffy, N., Bednarski, D. W., Schummer, M. and Haussler, D. (2000). Support vector machine classification and validation of cancer tissue samples using microarray expression data. Bioinformatics. 16: 906-914.

~~Gabor Miklos, G. L. and Maleszka, R. (2004). Microarray reality checks in the context of a complex disease. Nature Biotechnology 22: 615-621.~~

Gambaletta, D., Marchetti, A., Benedetti, L., Mercurio, A. M., Sacchi, A. and Falcioni, R. (2000). Cooperative signaling between alpha(6)beta(4) integrin and ErbB-2 receptor is required to promote phosphatidylinositol 3-kinase-dependent invasion. Journal of Biological Chemistry 275: 10604-10610.

Ganti, R., Skapek, S. X., Zhang, J., Fuller, C. E., Wu, J., Billups, C. A., Breitfeld, P. P., Dalton, J. D., Meyer, W. H. and Khoury, J. D. (2006). Expression and genomic status of EGFR and ErbB-2 in alveolar and embryonal rhabdomyosarcoma. Modern Pathology 19: 1213-1220.

Garber, J. E., Goldstein, A. M., Kantor, A. F., Dreyfus, M. G., Fraumeni, J. F., Jr. and Li, F. P. (1991). Follow-up study of twenty-four families with Li-Fraumeni syndrome. Cancer Research 51: 6094-6097.

~~Garcia, R. and Jove, R. (1998). Activation of STAT transcription factors in oncogenic tyrosine kinase signaling. Journal of Biomedical Science 5: 79-85.~~

Garzino-Demo, P., Carrozzo, M., Trusolino, L., Savoia, P., Gandolfo, S. and Marchisio, P. C. (1998). Altered expression of alpha 6 integrin subunit in oral squamous cell carcinoma and oral potentially malignant lesions. Oral Oncology 34: 204-10.

Gayed, I., Vu, T., Iyer, R., Johnson, M., Macapinlac, H., Swanston, N. and Podoloff, D. (2004). The role of 18F-FDG PET in staging and early prediction of response to therapy of recurrent gastrointestinal stromal tumors. Journal of Nuclear Medicine 45: 17-21.

~~432~~

~~Bibliography~~

~~Genter, M. B., Burman, D. M., Vijayakumar, S., Ebert, C. L. and Aronow, B. J. (2002). Genomic analysis of alachlor-induced oncogenesis in rat olfactory mucosa. Physiological Genomics 12: 35-45.~~

~~Gilcrease, M. Z. (2007). Integrin signaling in epithelial cells. Cancer Letters 247: 1-25.~~

~~Gillett, C. E., Springall, R. J., Barnes, D. M. and Hanby, A. M. (2000). Multiple tissue core arrays in histopathology research: a validation study. Journal of Pathology 192: 549-553.~~

Ginestier, C., Charafe-Jauffret, E., Bertucci, F., Eisinger, F., Geneix, J., Bechlian, D., Conte, N., Adelaide, J., Toiron, Y., Nguyen, C., Viens, P., Mozziconacci, M.-J., Houlgatte, R., Birnbaum, D. and Jacquemier, J. (2002). Distinct and Complementary Information Provided by Use of Tissue and DNA Microarrays in the Study of Breast Tumor Markers. American Journal of Pathology 161: 1223-1233.

Gisselsson, D., Jonson, T., Petersen, A., Strombeck, B., Dal Cin, P., Hoglund, M., Mitelman, F., Mertens, F. and Mandahl, N. (2001). Telomere dysfunction triggers extensive DNA fragmentation and evolution of complex chromosome abnormalities in human malignant tumors. Proceedings of the National Academy of Sciences of the United States of America 98: 12683-12688.

Golub, T. R., Slonim, D. K., Tamayo, P., Huard, C., Gaasenbeek, M., Mesirov, J. P., Coller, H., Loh, M. L., Downing, J. R., Caligiuri, M. A., Bloomfield, C. D. and Lander, E. S. (1999). Molecular classification of cancer: class discovery and class prediction by gene expression monitoring. Science 286: 531-537.

~~Gorelenkov, V., Antipov, A., Lejnine, S., Daraselia, N. and Yuryev, A. (2001). Set of Novel Tools for PCR Primer Design. Biotechniques. 31: 1326-1330.~~

Goto, Y., Kawauchi, S., Ihara, K., Ikemoto, K., Ohi, R., Kawai, S. and Sasaki, K. (2003). The prognosis in spindle-cell sarcoma depends on the expression of cyclin- dependent kinase inhibitor p27(Kip1) and cyclin E. Cancer Science. 94: 412-417.

Grego-Bessa, J., Diez, J., Timmerman, L. and de la Pompa, J. L. (2004). Notch and epithelial-mesenchyme transition in development and tumor progression: another turn of the screw. Cell Cycle 3: 718-721.

~~Griner, E. M. and Kazanietz, M. G. (2007). Protein kinase C and other diacylglycerol effectors in cancer. Nature Reviews. Cancer. 7: 281-294.~~

~~Grobmyer, S. R. and Brennan, M. F. (2003). Predictive variables detailing the recurrence rate of soft tissue sarcomas. Current Opinion in Oncology. 15: 319-326.~~

Gronwald, R. G., Adler, D. A., Kelly, J. D., Disteche, C. M. and Bowen-Pope, D. F. (1990). The human PDGF receptor alpha-subunit gene maps to chromosome 4 in close proximity to c-kit. Human Genetics 85: 383-385.

~~433~~

~~Bibliography~~

Grunhagen, D. J., Brunstein, F., ten Hagen, T. L. M., van Geel, A. N., de Wilt, H. W. and Eggermont, A. M. M. (2004). TNF-based isolated limb perfusion: A decade of experience with antivascular therapy in the management of locally advanced extremity soft tissue sarcomas. Targeting treatment of soft tissue sarcomas. J. Verweij and H. M. Pinedo. Dordrecht, Kluwer Academic Publishers: 65-80.

Guan, R. J., Ford, H. L., Fu, Y., Li, Y., Shaw, L. M. and Pardee, A. B. (2000). Drg-1 as a Differentiation-related, Putative Metastatic Suppressor Gene in Human Colon Cancer. Cancer Research 60: 749-755.

Guillou, L., Coindre, J., Gallagher, G., Terrier, P., Gebhard, S., de Saint Aubain Somerhausen, N., Michels, J., Jundt, G., Vince, D. R., Collin, F., Trassard, M., Le Doussal, V. and Benhattar, J. (2001). Detection of the synovial sarcoma translocation t(X;18) (SYT;SSX) in paraffin-embedded tissues using reverse transcriptase- polymerase chain reaction: a reliable and powerful diagnostic tool for pathologists. A molecular analysis of 221 mesenchymal tumors fixed in different fixatives. Human Pathology 32: 105-112.

Gunawan, B., Bergmann, F., Hoer, J., Langer, C., Schumpelick, V., Becker, H. and Fuzesi, L. (2002). Biological and clinical significance of cytogenetic abnormalities in low-risk and high-risk gastrointestinal stromal tumors. Human Pathology 33: 316-321.

Gunawan, B., Braun, S., Cortes, M. J., Bergmann, F., Karl, C. and Fuzesi, L. (1998). Characterization of a newly established endometrial stromal sarcoma cell line. International Journal of Cancer. 77: 424-8.

Gunther, T., Schneider-Stock, R., Hackel, C., Pross, M., Schulz, H.-U., Lippert, H. and Roessner, A. (2000). Telomerase Activity and Expression of hTRT and hTR in Gastrointestinal Stromal Tumors in Comparison with Extragastrointestinal Sarcomas. Clinical Cancer Research 6: 1811-1818.

Haas-Kogan, D. A., Prados, M. D., Tihan, T., Eberhard, D. A., Jelluma, N., Arvold, N. D., Baumber, R., Lamborn, K. R., Kapadia, A., Malec, M., Berger, M. S. and Stokoe, D. (2005). Epidermal growth factor receptor, protein kinase B/Akt, and glioma response to erlotinib. Journal of the National Cancer Institute 97: 880-887.

Hahn, S. A., Schutte, M., Hoque, A. T., Moskaluk, C. A., da Costa, L. T., Rozenblum, E., Weinstein, C. L., Fischer, A., Yeo, C. J., Hruban, R. H. and Kern, S. E. (1996). DPC4, a candidate tumor suppressor gene at human chromosome 18q21.1. Science 271: 350-353.

~~Hainaut, P. (1991, July 2004). IARC TP53 mutation database. R9. Retrieved 2 May 2005, 2005, from http://www-p53.iarc.fr/index.html.~~

Haley, J., Whittle, N., Bennet, P., Kinchington, D., Ullrich, A. and Waterfield, M. (1987). The human EGF receptor gene: structure of the 110 kb locus and identification of sequences regulating its transcription. Oncogene Research 1: 375-396.

~~434~~

~~Bibliography~~

Han, H., Bearss, D. J., Browne, L. W., Calaluce, R., Nagle, R. B. and Von Hoff, D. D. (2002). Identification of Differentially Expressed Genes in Pancreatic Cancer Cells Using cDNA Microarray. Cancer Research 62: 2890-2896.

Han, S.-W., Hwang, P. G., Chung, D. H., Kim, D.-W., Im, S.-A., Kim, Y. T., Kim, T.- Y., Heo, D. S., Bang, Y.-J. and Kim, N. K. (2005). Epidermal growth factor receptor (EGFR) downstream molecules as response predictive markers for gefitinib (Iressa®, ZD1839) in chemotherapy-resistant non-small cell lung cancer. International Journal of Cancer 113: 109-115.

~~Hanahan, D. and Folkman, J. (1996). Patterns and emerging mechanisms of the angiogenic switch during tumorigenesis. Cell 86: 353-364.~~

~~Hanahan, D. and Weinberg, R. A. (2000). The hallmarks of cancer. Cell 100: 57-70.~~

~~Hartmann, J. T. (2007). Systemic treatment options for patients with refractory adult- type sarcoma beyond anthracyclines. Anti Cancer Drugs 18: 245-254.~~

Hedenfalk, I., Duggan, D., Chen, Y., Radmacher, M., Bittner, M., Simon, R., Meltzer, P., Gusterson, B., Esteller, M., Raffeld, M., Yakhini, Z., Ben-Dor, A., Dougherty, E., Kononen, J., Bubendorf, L., Fehrle, W., Pittaluga, S., Gruvberger, S., Loman, N., Johannsson, O., Olsson, H., Wilfond, B., Sauter, G., Kallioniemi, O.-P., Borg, A. and Trent, J. (2001). Gene-Expression Profiles in Hereditary Breast Cancer. New England Journal of Medicine 344: 539-548.

Hegazy, M. A., Kotb, S. Z., Sakr, H., El Dosoky, E., Amer, T., Hegazi, R. A. and Farouk, O. (2007). Preoperative isolated limb infusion of Doxorubicin and external irradiation for limb-threatening soft tissue sarcomas. Annals of Surgical Oncology 14: 568-576.

~~Heid, C., Stevens, J., Livak, K. and Williams, P. (1996). Real time quantitative PCR. Genome Research 6: 986-994.~~

Heinrich, M. C., Corless, C. L., Demetri, G. D., Blanke, C. D., von Mehren, M., Joensuu, H., McGreevey, L. S., Chen, C.-J., Van den Abbeele, A. D., Druker, B. J., Kiese, B., Eisenberg, B., Roberts, P. J., Singer, S., Fletcher, C. D. M., Silberman, S., Dimitrijevic, S. and Fletcher, J. A. (2003a). Kinase Mutations and Imatinib Response in Patients With Metastatic Gastrointestinal Stromal Tumor. Journal of Clinical Oncology 21: 4342-4349.

Heinrich, M. C., Corless, C. L., Duensing, A., McGreevey, L., Chen, C. J., Joseph, N., Singer, S., Griffith, D. J., Haley, A., Town, A., Demetri, G. D., Fletcher, C. D. and Fletcher, J. A. (2003b). PDGFRA activating mutations in gastrointestinal stromal tumors. Science 299: 708-710.

Heinrich, M. C., Rubin, B. P., Longley, B. J. and Fletcher, J. A. (2002). Biology and genetic aspects of gastrointestinal stromal tumors: KIT activation and cytogenetic alterations. Human Pathology 33: 484-495.

~~435~~

~~Bibliography~~

~~Helman, L. J. and Meltzer, P. (2003). Mechanisms of sarcoma development. Nature Reviews Cancer 3: 685-694.~~

Henson, J. D., Hannay, J. A., McCarthy, S. W., Royds, J. A., Yeager, T. R., Robinson, R. A., Wharton, S. B., Jellinek, D. A., Arbuckle, S. M., Yoo, J., Robinson, B. G., Learoyd, D. L., Stalley, P. D., Bonar, S. F., Yu, D., Pollock, R. E. and Reddel, R. R. (2005). A robust assay for alternative lengthening of telomeres in tumors shows the significance of alternative lengthening of telomeres in sarcomas and astrocytomas. Clinical Cancer Research 11: 217-225.

~~Henson, J. D., Neumann, A. A., Yeager, T. R. and Reddel, R. R. (2002). Alternative lengthening of telomeres in mammalian cells. Oncogene 21: 598-610.~~

~~Hieken, T. J. and Das Gupta, T. K. (1996). Mutant p53 expression: a marker of diminished survival in well-differentiated soft tissue sarcoma. Clinical Cancer Research 2: 1391-1395.~~

Hill, D. A. M. D., O'Sullivan, M. J. M. D., Zhu, X. M. D., Vollmer, R. T. M. D., Humphrey, P. A. M. D. P. D., Dehner, L. P. M. D. and Pfeifer, J. D. M. D. P. D. (2002). Practical Application of Molecular Genetic Testing as an Aid to the Surgical Pathologic Diagnosis of Sarcomas: A Prospective Study. American Journal of Surgical Pathology August 26: 965-977.

Hiraga, H., Nojima, T., Abe, S., Yamashiro, K., Yamawaki, S., Kaneda, K. and Nagashima, K. (1997). Establishment of a new continuous clear cell sarcoma cell line. Morphological and cytogenetic characterization and detection of chimaeric EWS/ATF-1 transcripts. Virchows Archives 431: 45-51.

Hirota, S., Isozaki, K., Moriyama, Y., Hashimoto, K., Nishida, T., Ishiguro, S., Kawano, K., Hanada, M., Kurata, A., Takeda, M., Tunio, G. M., Matsuzawa, Y., Kanakura, Y., Shinomura, Y. and Kitamura, Y. (1998). Gain-of-Function Mutations of c-kit in Human Gastrointestinal Stromal Tumors. Science 279: 577-580.

Hirsch, F. R., Varella-Garcia, M., Bunn, P. A., Jr., Di Maria, M. V., Veve, R., Bremmes, R. M., Baron, A. E., Zeng, C. and Franklin, W. A. (2003). Epidermal growth factor receptor in non-small-cell lung carcinomas: correlation between gene copy number and protein expression and impact on prognosis. Journal of Clinical Oncology 21: 3798-3807.

Hisada, M., Garber, J. E., Fung, C. Y., Fraumeni, J. F., Jr. and Li, F. P. (1998). Multiple primary cancers in families with Li-Fraumeni syndrome. Journal of the National Cancer Institute. 90: 606-611.

~~Hoekman, K. and Pinedo, H. M. (2004). Angiogenesis: a potential target for therapy of soft tissue sarcomas. Cancer Treatment & Research 120: 169-180.~~

~~Hogendoorn, P. C., Collin, F., Daugaard, S., Dei Tos, A. P., Fisher, C., Schneider, U., Sciot, R., Pathology, Biology Subcommittee of the, E. S. T. and Bone Sarcoma, G.~~

~~436~~

~~Bibliography~~

~~(2004). Changing concepts in the pathological basis of soft tissue and bone sarcoma treatment. European Journal of Cancer 40: 1644-1654.~~

~~Holloway, A. J., van Laar, R. K., Tothill, R. W. and Bowtell, D. D. (2002). Options available--from start to finish--for obtaining data from DNA microarrays II. Nature Genetics. 32: 481-9.~~

~~Holzman, T. and Kolker, E. (2004). Statistical analysis of global gene expression data: Some practical considerations. Current Opinion in Biotechnology 15: 52-57.~~

~~Hood, J. D. and Cheresh, D. A. (2002). Role of integrins in cell invasion and migration. Nature Reviews Cancer 2: 91-100.~~

~~Hoos, A. and Cordon-Cardo, C. (2001). Tissue Microarray Profiling of Cancer Specimens and Cell Lines: Opportunities and Limitations. Laboratory Investigation 81: 1331-1338.~~

Hoos, A., Urist, M. J., Stojadinovic, A., Mastorides, S., Dudas, M. E., Leung, D. H., Kuo, D., Brennan, M. F., Lewis, J. J. and Cordon-Cardo, C. (2001). Validation of tissue microarrays for immunohistochemical profiling of cancer specimens using the example of human fibroblastic tumors. American Journal of Pathology 158: 1245-1251.

Horvai, A. E., Kramer, M. J. and O'Donnell, R. (2006). Beta-catenin nuclear expression correlates with cyclin D1 expression in primary and metastatic synovial sarcoma: a tissue microarray study. Archives of Pathology & Laboratory Medicine 130: 792-798.

Hsu, F. D., Nielsen, T. O., Alkushi, A., Dupuis, B., Huntsman, D., Liu, C. L., van de Rijn, M. and Gilks, C. B. (2002). Tissue microarrays are an effective quality assurance tool for diagnostic immunohistochemistry. Modern Pathology. 15: 1374-1380.

Hsu, S. M., Rohol, P. J. M., Ko, Y. H. and Lok, M. S. (1989). Malignant fibrous histiocytoma is not related to histiocytes: Consistent phenotypic expression of CD13, CD10, and vimentin in four malignant fibrous histiocytoma cell lines. Cancer Journal 2: 423-429.

Hu, M., Nicolson, G. L., Trent, J. C., 2nd, Yu, D., Zhang, L., Lang, A., Killary, A., Ellis, L. M., Bucana, C. D. and Pollock, R. E. (2002). Characterization of 11 human sarcoma cell strains: evaluation of cytogenetics, tumorigenicity, metastasis, and production of angiogenic factors. Cancer. 95: 1569-76.

~~Hunt, K. K., Vorburger, S. A. and Cormier, J. N. (2002). Surgical Approaches to Extremity Sarcoma. Atlas of Clinical Oncology: Soft Tissue Sarcomas. R. E. Pollock. Philadelphia, BC Decker Inc: 89-109.~~

~~Hunter, T. (1995). Protein kinases and phosphatases: the yin and yang of protein phosphorylation and signaling. Cell 80: 225-236.~~

~~Hynes, N. E., Lane, H. A., Hynes, N. E. and Lane, H. A. (2005). ERBB receptors and cancer: the complexity of targeted inhibitors. Nature Reviews Cancer 5: 341-54.~~

~~437~~

~~Bibliography~~

~~Ihaka, R. and Gentleman, R. (1996). R: A language for data analysis and graphics. Journal of Computational and Graphical Statistics 5: 299-314.~~

~~Ihle, J. N. (1995). Cytokine receptor signalling. Nature 377: 591-594.~~

Iizuka, N., Oka, M., Yamada-Okabe, H., Nishida, M., Maeda, Y., Mori, N., Takao, T., Tamesa, T., Tangoku, A., Tabuchi, H., Hamada, K., Nakayama, H., Ishitsuka, H., Miyamoto, T., Hirabayashi, A., Uchimura, S. and Hamamoto, Y. (2003). Oligonucleotide microarray for prediction of early intrahepatic recurrence of hepatocellular carcinoma after curative resection. Lancet 361: 923-929.

Irizarry, R. A., Warren, D., Spencer, F., Kim, I. F., Biswal, S., Frank, B. C., Gabrielson, E., Garcia, J. G., Geoghegan, J., Germino, G., Griffin, C., Hilmer, S. C., Hoffman, E., Jedlicka, A. E., Kawasaki, E., Martinez-Murillo, F., Morsberger, L., Lee, H., Petersen, D., Quackenbush, J., Scott, A., Wilson, M., Yang, Y., Ye, S. Q. and Yu, W. (2005). Multiple-laboratory comparison of microarray platforms. Nature Methods 2: 345-350.

Isozaki, K., Terris, B., Belghiti, J., Schiffmann, S., Hirota, S. and Vanderwinden, J. M. (2000). Germline-activating mutation in the kinase domain of KIT gene in familial gastrointestinal stromal tumors. American Journal of Pathology 157: 1581-1585.

~~Israel-Mardirosian, N. and Adler, L. P. (2003). Positron emission tomography of soft tissue sarcomas. Current Opinion in Oncology. 15: 327-330.~~

Jacobs, T. W., Prioleau, J. E., Stillman, I. E. and Schnitt, S. J. (1996). Loss of Tumor Marker-Immunostaining Intensity on Stored Paraffin Slides of Breast Cancer. Journal of the National Cancer Institute. 88: 1054-1059.

Jarvinen, A. K., Hautaniemi, S., Edgren, H., Auvinen, P., Saarela, J., Kallioniemi, O. P. and Monni, O. (2004). Are data from different gene expression microarray platforms comparable? Genomics 83: 1164-1168.

~~Jin, H. and Varner, J. (2004). Integrins: roles in cancer development and as treatment targets. British Journal of Cancer 90: 561-565.~~

Johnson, J. E., Gettings, E. J., Schwalm, J., Pei, J., Testa, J. R., Litwin, S., von Mehren, M. and Broccoli, D. (2007). Whole-Genome Profiling in Liposarcomas Reveals Genetic Alterations Common to Specific Telomere Maintenance Mechanisms. Cancer Research 67: 9221-9228.

Johnson, J. E., Varkonyi, R. J., Schwalm, J., Cragle, R., Klein-Szanto, A., Patchefsky, A., Cukierman, E., von Mehren, M. and Broccoli, D. (2005). Multiple Mechanisms of Telomere Maintenance Exist in Liposarcomas. Clinical Cancer Research 11: 5347- 5355.

Jorissen, R. N., Walker, F., Pouliot, N., Garrett, T. P., Ward, C. W. and Burgess, A. W. (2003). Epidermal growth factor receptor: mechanisms of activation and signalling. Experimental Cell Research 284: 31-53.

~~438~~

~~Bibliography~~

Jourdan, F., Sebbagh, N., Comperat, E., Mourra, N., Flahault, A., Olschwang, S., Duval, A., Hamelin, R. and Flejou, J.-F. o. (2003). Tissue microarray technology: validation in colorectal carcinoma and analysis of p53, hMLH1, and hMSH2 immunohistochemical expression. Virchows Archives 443: 115-121.

Juric, D., Lacayo, N. J., Ramsey, M. C., Racevskis, J., Wiernik, P. H., Rowe, J. M., Goldstone, A. H., O'Dwyer, P. J., Paietta, E. and Sikic, B. I. (2007). Differential Gene Expression Patterns and Interaction Networks in BCR-ABL-Positive and -Negative Adult Acute Lymphoblastic Leukemias. Journal of Clinical Oncology 25: 1341-1349.

Kalaydjieva, L., Gresham, D., Gooding, R., Heather, L., Baas, F., de Jonge, R., Blechschmidt, K., Angelicheva, D., Chandler, D., Worsley, P., Rosenthal, A., King, R. H. and Thomas, P. K. (2000). N-myc downstream-regulated gene 1 is mutated in hereditary motor and sensory neuropathy-Lom. American Journal of Human Genetics 67: 47-58.

~~Kallioniemi, O.-P., Wagner, U., Kononen, J. and Sauter, G. (2001). Tissue microarray technology for high-throughput molecular profiling of cancer. Human Molecular Genetics. 10: 657-662.~~

Kalmes, A., Vesti, B. R., Daum, G., Abraham, J. A. and Clowes, A. W. (2000). Heparin blockade of thrombin-induced smooth muscle cell migration involves inhibition of epidermal growth factor (EGF) receptor transactivation by heparin-binding EGF-like growth factor. Circulation Research 87: 92-98.

Kanoe, H., Nakayama, T., Hosaka, T., Murakami, H., Yamamoto, H., Nakashima, Y., Tsuboyama, T., Nakamura, T., Ron, D., Sasaki, M. S. and Toguchida, J. (1999). Characteristics of genomic breakpoints in TLS-CHOP translocations in liposarcomas suggest the involvement of Translin and topoisomerase II in the process of translocation. Oncogene 18: 721-729.

Kanoe, H., Nakayama, T., Murakami, H., Hosaka, T., Yamamoto, H., Nakashima, Y., Tsuboyama, T., Nakamura, T., Sasaki, M. S. and Toguchida, J. (1998). Amplification of the CDK4 gene in sarcomas: tumor specificity and relationship with the RB gene mutation. Anticancer Research. 18: 2317-2321.

Kappler, M., Köhler, T., Kampf, C., Diestelkötter, P., Würl, P., Schmitz, M., Bartel, F., Lautenschläger, C., Rieber, E. P., Schmidt, H., Bache, M., Taubert, H. and Meye, A. (2001). Increased survivin transcript levels: An independent negative predictor of survival in soft tissue sarcoma patients. International Journal of Cancer 95: 360-363.

Kappler, M., Kotzsch, M., Bartel, F., Fussel, S., Lautenschlager, C., Schmidt, U., Wurl, P., Bache, M., Schmidt, H., Taubert, H. and Meye, A. (2003). Elevated expression level of survivin protein in soft-tissue sarcomas is a strong independent predictor of survival. Clinical Cancer Research. 9: 1098-1104.

~~Kasper, B., Gil, T., D'Hondt, V., Gebhart, M. and Awada, A. (2007). Novel treatment strategies for soft tissue sarcoma. Critical Reviews in Oncology Hematology 62: 9-15.~~

~~439~~

~~Bibliography~~

Kauffman, E. C., Robinson, V. L., Stadler, W. M., Sokoloff, M. H. and Rinker- Schaeffer, C. W. (2003). Metastasis suppression: the evolving role of metastasis suppressor genes for regulating cancer cell growth at the secondary site. Journal of Urology 169: 1122-1133.

Kawaguchi, K., Oda, Y., Saito, T., Yamamoto, H., Tamiya, S., Takahira, T., Miyajima, K., Iwamoto, Y. and Tsuneyoshi, M. (2003). Mechanisms of inactivation of the p16INK4a gene in leiomyosarcoma of soft tissue: decreased p16 expression correlates with promoter methylation and poor prognosis. Journal of Pathology. 201: 487-95.

Kawaguchi, S., Wada, T., Ida, K., Sato, Y., Nagoya, S., Tsukahara, T., Kimura, S., Sahara, H., Ikeda, H., Shimozawa, K., Asanuma, H., Torigoe, T., Hiraga, H., Ishii, T., Tatezaki, S. I., Sato, N. and Yamashita, T. (2005). Phase I vaccination trial of SYT-SSX junction peptide in patients with disseminated synovial sarcoma. Journal of Translational Medicine 3: 1.

~~Kaytan, E., Yaman, F., Cosar, R., Eralp, Y., Saip, P. and Darendeliler, E. (2003). Prognostic factors in localized soft-tissue sarcomas. American Journal of Clinical Oncology. 26: 411-415.~~

~~Kerr, M. K. (2003). Experimental design to make the most of microarray studies. Methods in Molecular Biology 224: 137-147.~~

~~Kerr, M. K. and Churchill, G. A. (2001a). Experimental design for gene expression microarrays. Biostatistics 2: 183-201.~~

~~Kerr, M. K. and Churchill, G. A. (2001b). Statistical design and the analysis of gene expression microarray data. Genetical Research. 77: 123-128.~~

~~Kerr, M. K., Martin, M. and Churchill, G. A. (2000). Analysis of variance for gene expression microarray data. Journal of Computational Biology. 7: 819-37.~~

Kersting, C., Packeisen, J., Leidinger, B., Brandt, B., von Wasielewski, R., Winkelmann, W., van Diest, P. J., Gosheger, G. and Buerger, H. (2006). Pitfalls in immunohistochemical assessment of EGFR expression in soft tissue sarcomas. Journal of Clinical Pathology 59: 585-590.

Khan, J., Bittner, M. L., Saal, L. H., Teichmann, U., Azorsa, D. O., Gooden, G. C., Pavan, W. J., Trent, J. M. and Meltzer, P. S. (1999). cDNA microarrays detect activation of a myogenic transcription program by the PAX3-FKHR fusion oncogene. Proceedings of the National Academy of Sciences of the United States of America 96: 13264-13269.

Khan, J., Simon, R., Bittner, M., Chen, Y., Leighton, S. B., Pohida, T., Smith, P. D., Jiang, Y., Gooden, G. C., Trent, J. M. and Meltzer, P. S. (1998). Gene expression profiling of alveolar rhabdomyosarcoma with cDNA microarrays. Cancer Research 58: 5009-5013.

~~440~~

~~Bibliography~~

Khan, J., Wei, J. S., Ringner, M., Saal, L. H., Ladanyi, M., Westermann, F., Berthold, F., Schwab, M., Antonescu, C. R., Peterson, C. and Meltzer, P. S. (2001). Classification and diagnostic prediction of cancers using gene expression profiling and artificial neural networks. Nature Medicine 7: 673-679.

Kikuchi, T., Daigo, Y., Katagiri, T., Tsunoda, T., Okada, K., Kakiuchi, S., Zembutsu, H., Furukawa, Y., Kawamura, M., Kobayashi, K., Imai, K. and Nakamura, Y. (2003). Expression profiles of non-small cell lung cancers on cDNA microarrays: identification of genes for prediction of lymph-node metastasis and sensitivity to anti-cancer drugs. Oncogene 22: 2192-2205.

Kilian, A., Bowtell, D., Abud, H., Hime, G., Venter, D., Keese, P., Duncan, E., Reddel, R. and Jefferson, R. (1997). Isolation of a candidate human telomerase catalytic subunit gene, which reveals complex splicing patterns in different cell types. Human Molecular Genetics. 6: 2011-2019.

Kim, B., Bang, S., Lee, S., Kim, S., Jung, Y., Lee, C., Choi, K., Lee, S.-G., Lee, K., Lee, Y., Kim, S.-S., Yeom, Y.-I., Kim, Y.-S., Yoo, H.-S., Song, K. and Lee, I. (2003). Expression Profiling and Subtype-Specific Expression of Stomach Cancer. Cancer Research 63: 8248-8255.

Kim, M. H., Jung, M. A., Hwang, Y. S., Jeong, M., Kim, S. M., Ahn, S. J., Shin, B. A., Ahn, B. W. and Jung, Y. D. (2004). Regulation of urokinase plasminogen activator by epigallocatechin-3- gallate in human fibrosarcoma cells. European Journal of Pharmacology 487: 1-6.

Kim, N. W., Piatyszek, M. A., Prowse, K. R., Harley, C. B., West, M. D., Ho, P. L., Coviello, G. M., Wright, W. E., Weinrich, S. L. and Shay, J. W. (1994). Specific association of human telomerase activity with immortal cells and cancer. Science 266: 2011-2015.

Kim, S., Bell, K., Mousa, S. A. and Varner, J. A. (2000). Regulation of Angiogenesis in Vivo by Ligation of Integrin α5β1 with the Central Cell-Binding Domain of Fibronectin. American Journal of Pathology 156: 1345-1362.

~~Kim, S. H., Cho, N. H., Tallini, G., Dudas, M., Lewis, J. J. and Cordon-Cardo, C. (2001). Prognostic role of cyclin D1 in retroperitoneal sarcomas. Cancer 91: 428-434.~~

Kindblom, L. G., Remotti, H. E., Aldenborg, F. and Meis-Kindblom, J. M. (1998). Gastrointestinal pacemaker cell tumor (GIPACT): gastrointestinal stromal tumors show phenotypic characteristics of the interstitial cells of Cajal. American Journal of Pathology 152: 1259-1269.

~~King, H. C. and Sinha, A. A. (2001). Gene expression profile analysis by DNA microarrays: promise and pitfalls. Journal of the American Medical Association 286: 2280-2288.~~

~~Kiyozuka, Y., Nakagawa, H., Uemura, Y., Senzaki, H., Yamamoto, A., Noguchi, T., Mizuta, H., Nakanishi, K., Nakano, S. and Tsubura, A. (2001). Novel cell lines~~

~~441~~

~~Bibliography~~

~~established from a human myxoid malignant fibrous histiocytoma arising in the uterus. Cancer Genetics and Cytogenetics 127: 7-15.~~

~~Knight, W. A., 3rd, Murphy, W. K. and Gottlieb, J. A. (1973). Neurofibromatosis associated with malignant neurofibromas. Archives of Dermatology 107: 747-750.~~

~~Knudsen, K. E., Diehl, J. A., Haiman, C. A. and Knudsen, E. S. (2006). Cyclin D1: polymorphism, aberrant splicing and cancer risk. Oncogene 25: 1620-1628.~~

Koea, J. B., Leung, D., Lewis, J. J. and Brennan, M. F. (2003). Histopathologic type: an independent prognostic factor in primary soft tissue sarcoma of the extremity? Annals of Surgical Oncology. 10: 432-40.

Kohler, T., Wurl, P., Meye, A., Lautenschlager, C., Bartel, F., Borchert, S., Bache, M., Schmidt, H., Holzhausen, H. J. and Taubert, H. (2002). High bad and bcl-xL gene expression and combined bad, bcl-xL, bax and bcl-2 mRNA levels: molecular predictors for survival of stage 2 soft tissue sarcoma patients. Anticancer Research 22: 1553-1559.

Komdeur, R., Hoekstra, H. J., van den Berg, E., Molenaar, W. M., Pras, E., de Vries, E. G. and van der Graaf, W. T. (2002). Metastasis in soft tissue sarcomas: prognostic criteria and treatment perspectives. Cancer Metastasis Reviews. 21: 167-83.

Konomoto, T., Fukuda, T., Hayashi, K., Kumazawa, J. and Tsuneyoshi, M. (1998). Leiomyosarcoma in soft tissue: examination of p53 status and cell proliferating factors in different locations. Human Pathology. 29: 74-81.

Kononen, J., Bubendorf, L., Kallioniemi, A., Barlund, M., Schraml, P., Leighton, S., Torhorst, J., Mihatsch, M. J., Sauter, G. and Kallioniemi, O. P. (1998). Tissue microarrays for high-throughput molecular profiling of tumor specimens. Nature Medicine 4: 844-847.

~~Kosciolek, B. A., Kalantidis, K., Tabler, M. and Rowley, P. T. (2003). Inhibition of Telomerase Activity in Human Cancer Cells by RNA Interference. Molecular Cancer Therapeutics 2: 209-216.~~

~~Kothapalli, R., Yoder, S. J., Mane, S. and Loughran, T. P., Jr. (2002). Microarray results: how accurate are they? BMC Bioinformatics 3: 22.~~

~~Kuo, W. P., Jenssen, T., Butte, A. J., Ohno-Machado, L. and Kohane, I. S. (2002). Analysis of matched mRNA measurements from two different microarray technologies. Bioinformatics 18: 405-412.~~

Kuttesch, J. F., Jr., Wexler, L. H., Marcus, R. B., Fairclough, D., Weaver-McClure, L., White, M., Mao, L., Delaney, T. F., Pratt, C. B., Horowitz, M. E. and Kun, L. E. (1996). Second malignancies after Ewing's sarcoma: radiation dose-dependency of secondary sarcomas. Journal of Clinical Oncology 14: 2818-2825.

~~442~~

~~Bibliography~~

Lachat, P., Shaw, P., Gebhard, S., van Belzen, N., Chaubert, P. and Bosman, F. T. (2002). Expression of NDRG1, a differentiation-related gene, in human tissues. Histochemistry and Cell Biology 118: 399 - 408.

~~Ladanyi, M. and Bridge, J. A. (2000). Contribution of molecular genetic data to the classification of sarcomas. Human Pathology 31: 532-538.~~

~~Lai, E. C. (2004). Notch signaling: control of cell communication and cell fate. Development 131: 965-973.~~

Laird, A. D., Vajkoczy, P., Shawver, L. K., Thurnher, A., Liang, C., Mohammadi, M., Schlessinger, J., Ullrich, A., Hubbard, S. R., Blake, R. A., Fong, T. A. T., Strawn, L. M., Sun, L., Tang, C., Hawtin, R., Tang, F., Shenoy, N., Hirth, K. P., McMahon, G. and Cherrington, J. M. (2000). SU6668 Is a Potent Antiangiogenic and Antitumor Agent That Induces Regression of Established Tumors. Cancer Research 60: 4152-4160.

~~Lamb, J. and Ewen, M. E. (2003). Cyclin D1 and molecular chaperones: implications for tumorigenesis. Cell Cycle 2: 525-527.~~

Langer, C., Gunawan, B., Schüler, P., Huber, W., Füzesi, L. and Becker, H. (2003). Prognostic factors influencing surgical management and outcome of gastrointestinal stromal tumours. British Journal of Surgery 90: 332-339.

~~Lappohn, R. E., Burger, H. G., Bouma, J., Bangah, M., Krans, M. and de Bruijn, H. W. (1989). Inhibin as a marker for granulosa-cell tumors. New England Journal of Medicine 321: 790-793.~~

~~Larkin, J. E., Frank, B. C., Gavras, H., Sultana, R. and Quackenbush, J. (2005). Independence and reproducibility across microarray platforms. Nature Methods 2: 337- 344.~~

Laskar, S., Bahl, G., Puri, A., Agarwal, M. G., Muckaden, M., Patil, N., Jambhekar, N., Gupta, S., Deshpande, D. D., Shrivastava, S. K. and Dinshaw, K. A. (2007). Perioperative interstitial brachytherapy for soft tissue sarcomas: prognostic factors and long-term results of 155 patients. Annals of Surgical Oncology 14: 560-567.

Lasota, J., Jasinski, M., Sarlomo-Rikala, M. and Miettinen, M. (1999). Mutations in exon 11 of c-Kit occur preferentially in malignant versus benign gastrointestinal stromal tumors and do not occur in leiomyomas or leiomyosarcomas. American Journal of Pathology 154: 53-60.

Lee, J. H., Park, S. R., Chay, K. O., Seo, Y. W., Kook, H., Ahn, K. Y., Kim, Y. J. and Kim, K. K. (2004). KAI1 COOH-terminal interacting tetraspanin (KITENIN), a member of the tetraspanin family, interacts with KAI1, a tumor metastasis suppressor, and enhances metastasis of cancer. Cancer Research 64: 4235-4243.

Lee, Y. F., John, M., Edwards, S., Clark, J., Flohr, P., Maillard, K., Edema, M., Baker, L., Mangham, D. C., Grimer, R., Wooster, R., Thomas, J. M., Fisher, C., Judson, I. and Cooper, C. S. (2003). Molecular classification of synovial sarcomas, leiomyosarcomas

~~443~~

~~Bibliography~~

~~and malignant fibrous histiocytomas by gene expression profiling. British Journal of Cancer 88: 510-515.~~

Lee, Y. F., John, M., Falconer, A., Edwards, S., Clark, J., Flohr, P., Roe, T., Wang, R., Shipley, J., Grimer, R. J., Mangham, D. C., Thomas, J. M., Fisher, C., Judson, I. and Cooper, C. S. (2004). A gene expression signature associated with metastatic outcome in human leiomyosarcomas. Cancer Research 64: 7201-4.

Leman, E. S., Madigan, M. C., Brunagel, G., Takaha, N., Coffey, D. S. and Getzenberg, R. H. (2003). Nuclear matrix localization of high mobility group protein I(Y) in a transgenic mouse model for prostate cancer. Journal of Cellular Biochemistry 88: 599- 608.

Lévy, P., Vidaud, D., Leroy, K., Laurendeau, I., Wechsler, J., Bolasco, G., Parfait, B., Wolkenstein, P., Vidaud, M. and Bièche, I. (2004). Molecular profiling of malignant peripheral nerve sheath tumors associated with neurofibromatosis type 1, based on large-scale real-time RT-PCR. Molecular Cancer 3.

~~Li, F. P. and Fraumeni, J. F., Jr. (1969a). Rhabdomyosarcoma in children; epidemiologic study and identification of a cancer family syndrome. J Natl Cancer Inst 43: 1365-1373.~~

~~Li, F. P. and Fraumeni, J. F., Jr. (1969b). Soft tissue sarcomas, breast cancer and other neoplasms: a family syndrome? Ann Int Med 71: 747-752.~~

Li, J. and Kretzner, L. (2003). The growth-inhibitory Ndrg1 gene is a Myc negative target in human neuroblastomas and other cell types with overexpressed N- or c-myc. Molecular and Cellular Biochemistry 250: 91 - 105.

~~Li, J. L. and Harris, A. (2005). Notch signaling from tumor cells: A new mechanism of angiogenesis. Cancer Cell 8: 1-3.~~

Liesveld, J. L., Rush, S., Kempski, M. C., Turner, A. R., Brennan, J. K., Gasson, J. C. and Abboud, C. N. (1993). Phenotypic characterization of the human fibrous histiocytoma giant cell tumor (GCT) cell line and its cytokine repertoire. Experimental Hematology 21: 1342-1352.

Lindvall, C., Hou, M., Komurasaki, T., Zheng, C., Henriksson, M., Sedivy, J. M., Bjorkholm, M., Teh, B. T., Nordenskjold, M. and Xu, D. (2003). Molecular Characterization of Human Telomerase Reverse Transcriptase-immortalized Human Fibroblasts by Gene Expression Profiling: Activation of the Epiregulin Gene. Cancer Research 63: 1743-1747.

~~Liu, J. H., Wei, S., Burnette, P. K., Gamero, A. M., Hutton, M. and Djeu, J. Y. (1999). Functional association of TGF-beta receptor II with cyclin B. Oncogene. 18: 269-275.~~

~~Lotze, M. T. and Tracey, K. J. (2005). High-mobility group box 1 protein (HMGB1): nuclear weapon in the immune arsenal. Nature Reviews Immunology. 5: 331-342.~~

~~444~~

~~Bibliography~~

Loughrey, M. B., Waring, P. M., Dobrovic, A., Demetri, G., Kovalenko, S. and McArthur, G. (2006). Polyclonal resistance in gastrointestinal stromal tumor treated with sequential kinase inhibitors. Clinical Cancer Research 12: 6205-6206.

Ludwig, A., Saretzki, G., Holm, P. S., Tiemann, F., Lorenz, M., Emrich, T., Harley, C. B. and von Zglinicki, T. (2001). Ribozyme cleavage of telomerase mRNA sensitizes breast epithelial cells to inhibitors of topoisomerase. Cancer Research 61: 3053-61.

~~Luttrell, L. M., Daaka, Y. and Lefkowitz, R. J. (1999). Regulation of tyrosine kinase cascades by G-protein-coupled receptors. Current Opinion in Cell Biology 11: 177-183.~~

Lux, M. L., Rubin, B. P., Biase, T. L., Chen, C. J., Maclure, T., Demetri, G., Xiao, S., Singer, S., Fletcher, C. D. and Fletcher, J. A. (2000). KIT extracellular and kinase domain mutations in gastrointestinal stromal tumors. American Journal of Pathology 156: 791-5.

Lynch, H. T., Brand, R. E., Hogg, D., Deters, C. A., Fusaro, R. M., Lynch, J. F., Liu, L., Knezetic, J., Lassam, N. J., Goggins, M. and Kern, S. (2002). Phenotypic variation in eight extended CDKN2A germline mutation familial atypical multiple mole melanoma- pancreatic carcinoma-prone families: the familial atypical mole melanoma-pancreatic carcinoma syndrome. Cancer 94: 84-96.

~~Lynch, H. T., Deters, C. A., Hogg, D., Lynch, J. F., Kinarsky, Y. and Gatalica, Z. (2003). Familial sarcoma: challenging pedigrees. Cancer. 98: 1947-1957.~~

Lynch, T. J., Bell, D. W., Sordella, R., Gurubhagavatula, S., Okimoto, R. A., Brannigan, B. W., Harris, P. L., Haserlat, S. M., Supko, J. G., Haluska, F. G., Louis, D. N., Christiani, D. C., Settleman, J. and Haber, D. A. (2004). Activating Mutations in the Epidermal Growth Factor Receptor Underlying Responsiveness of Non-Small-Cell Lung Cancer to Gefitinib. New England Journal of Medicine 350: 2129-2139.

Ma, X. J., Salunga, R., Tuggle, J. T., Gaudet, J., Enright, E., McQuary, P., Payette, T., Pistone, M., Stecker, K., Zhang, B. M., Zhou, Y. X., Varnholt, H., Smith, B., Gadd, M., Chatfield, E., Kessler, J., Baer, T. M., Erlander, M. G. and Sgroi, D. C. (2003). Gene expression profiles of human breast cancer progression. Proceedings of the National Academy of Sciences of the United States of America. 100: 5974-5979.

MacGeoch, C., Turner, G., Bobrow, L. G., Barnes, D. M., Bishop, D. T. and Spurr, N. K. (1995). Heterogeneity in Li-Fraumeni families: p53 mutation analysis and immunohistochemical staining. Journal of Medical Genetics 32: 186-90.

Mack, L. A., Crowe, P. J., Yang, J. L., Schachar, N. S., Morris, D. G., Kurien, E. C., Temple, C. L., Lindsay, R. L., Magi, E., DeHaas, W. G. and Temple, W. J. (2005). Preoperative chemoradiotherapy (modified Eilber protocol) provides maximum local control and minimal morbidity in patients with soft tissue sarcoma. Annals of Surgical Oncology 12: 646-653.

~~445~~

~~Bibliography~~

Maclean, K., Rogan, E. M., Whitaker, N. J., Chang, C. M., Rowe, P. B., Dalla-Pozza, L., Symonds, G. and Reddel, R. R. (1994). In vitro transformation of Li-Fraumeni syndrome fibroblasts by SV40 large T antigen mutants. Oncogene 9: 719-725.

Maio, M., Coral, S., Sigalotti, L., Elisei, R., Romei, C., Rossi, G., Cortini, E., Colizzi, F., Fenzi, G., Altomonte, M., Pinchera, A. and Vitale, M. (2003). Analysis of Cancer/Testis Antigens in Sporadic Medullary Thyroid Carcinoma: Expression and Humoral Response to NY-ESO-1. Journal of Clinical Endocrinology and Metabolism 88: 748-754.

Mairal, A., Chibon, F., Rousselet, A., Couturier, J., Terrier, P. and Aurias, A. (2000). Establishment of a Human Malignant Fibrous Histiocytoma Cell Line, COMA: Characterization by Conventional Cytogenetics, Comparative Genomic Hybridization, and Multiplex Fluorescence In Situ Hybridization. Cancer Genetics and Cytogenetics 121: 117-123.

Mairal, A., Terrier, P., Chibon, F., Sastre, X., Lecesne, A. and Aurias, A. (1999). Loss of chromosome 13 is the most frequent genomic imbalance in malignant fibrous histiocytomas. A comparative genomic hybridization analysis of a series of 30 cases. Cancer Genetics & Cytogenetics 111: 134-138.

Malkin, D., Li, F. P., Strong, L. C., Fraumeni, J. F., Jr., Nelson, C. E., Kim, D. H., Kassel, J., Gryka, M. A., Bischoff, F. Z., Tainsky, M. A. and Friend, S. H. (1990). Germ line p53 mutations in a familial syndrome of breast cancer, sarcomas, and other neoplasms. Science 250: 1233-1238.

Mandahl, N., Fletcher, C. D., Dal Cin, P., De Wever, I., Mertens, F., Mitelman, F., Rosai, J., Rydholm, A., Sciot, R., Tallini, G., Van Den Berghe, H., Vanni, R. and Willen, H. (2000). Comparative cytogenetic study of spindle cell and pleomorphic leiomyosarcomas of soft tissues: a report from the CHAMP Study Group. Cancer Genetics & Cytogenetics 116: 66-73.

~~Mandahl, N., Heim, S., Arheden, K., Rydholm, A., Willen, H. and Mitelman, F. (1988). Rings, dicentrics, and telomeric association in histiocytomas. Cancer Genetics & Cytogenetics 30: 23-33.~~

~~Mandahl, N., Mertens, F., Panagopoulos, I. and Knuutila, S. (2004). Genetic characterization of bone and soft tissue tumors. Acta Orthopaedica Scandinavica. Supplementum. 75: 21-28.~~

~~Markhede, G., Angervall, L. and Stener, B. (1982). A multivariate analysis of the prognosis after surgical treatment of malignant soft-tissue tumors. Cancer 49: 1721- 1733.~~

~~Marshall, C. J. (1995). Specificity of receptor tyrosine kinase signaling: transient versus sustained extracellular signal-regulated kinase activation. Cell 80: 179-185.~~

~~Masters, J. W. and Palsson, B. O., Eds. (1999). Cancer Cell Lines Part 1. Human Cell Culture. New York, Kluwer Academic.~~

~~446~~

~~Bibliography~~

~~Matzuk, M. M., Kumar, T. R. and Bradley, A. (1995). Different phenotypes for mice deficient in either activins or activin receptor type II. Nature 374: 356-360.~~

~~McCormick, F. (2001). Cancer gene therapy: fringe or cutting edge? Nature Reviews Cancer 1: 130-141.~~

McMullen, M. E., Bryant, P. W., Glembotski, C. C., Vincent, P. A. and Pumiglia, K. M. (2005). Activation of p38 Has Opposing Effects on the Proliferation and Migration of Endothelial Cells. Journal of Biological Chemistry 280: 20995-21003.

Mechtersheimer, G., Otano-Joos, M., Ohl, S., Benner, A., Lehnert, T., Willeke, F., Moller, P., Otto, H. F., Lichter, P. and Joos, S. (1999). Analysis of chromosomal imbalances in sporadic and NF1-associated peripheral nerve sheath tumors by comparative genomic hybridization. Genes, Chromosomes & Cancer 25: 362-369.

Medina-Franco, H., Ramos-De la Medina, A., Cortes-Gonzalez, R., Baquera, J., Angeles-Angeles, A., Urist, M. M., Eltoum, I. E. and Heslin, M. J. (2003). Expression of p53 and proliferation index as prognostic factors in gastrointestinal sarcomas. Annals of Surgical Oncology. 10: 190-195.

Mendel, D. B., Laird, A. D., Xin, X., Louie, S. G., Christensen, J. G., Li, G., Schreck, R. E., Abrams, T. J., Ngai, T. J., Lee, L. B., Murray, L. J., Carver, J., Chan, E., Moss, K. G., Haznedar, J. O., Sukbuntherng, J., Blake, R. A., Sun, L., Tang, C., Miller, T., Shirazian, S., McMahon, G. and Cherrington, J. M. (2003). In Vivo Antitumor Activity of SU11248, a Novel Tyrosine Kinase Inhibitor Targeting Vascular Endothelial Growth Factor and Platelet-derived Growth Factor Receptors: Determination of a Pharmacokinetic/Pharmacodynamic Relationship. Clinical Cancer Research 9: 327- 337.

~~Mendelsohn, J. and Baselga, J. (2003). Status of epidermal growth factor receptor antagonists in the biology and treatment of cancer. Journal of Clinical Oncology 21: 2787-2799.~~

Merimsky, O., Issakov, J., Schwartz, I., Dadia, S., Kollender, Y., Bickels, J., Inbar, M. and Meller, I. (2002). Lack of ErbB-2 oncogene product overexpression in soft tissue sarcomas. Acta Oncologica. 41: 366-8.

Mertens, F., Fletcher, C. D., Dal Cin, P., De Wever, I., Mandahl, N., Mitelman, F., Rosai, J., Rydholm, A., Sciot, R., Tallini, G., Van den Berghe, H., Vanni, R. and Willen, H. (1998). Cytogenetic analysis of 46 pleomorphic soft tissue sarcomas and correlation with morphologic and clinical features: a report of the CHAMP Study Group. Chromosomes and Morphology. Genes, Chromosomes & Cancer 22: 16-25.

Mertens, F., Stromberg, U., Mandahl, N., Cin, P. D., De Wever, I., Fletcher, C. D. M., Mitelman, F., Rosai, J., Rydholm, A., Sciot, R., Tallini, G., Van den Berghe, H., Vanni, R. and Willen, H. (2002). Prognostically important chromosomal aberrations in soft issue sarcomas: a report of the Chromosomes and Morphology (CHAMP) Study Group. Cancer Research 62: 3980-3984.

~~447~~

~~Bibliography~~

~~Michiels, S., Koscielny, S. and Hill, C. (2005). Prediction of cancer outcome with microarrays: a multiple random validation strategy. Lancet 365: 488-492.~~

~~Miller, L. D., Long, P. M., Wong, L., Mukherjee, S., McShane, L. M. and Liu, E. T. (2002). Optimal gene expression analysis by microarrays. Cancer Cell 2: 353-361.~~

Moch, H., Schraml, P., Bubendorf, L., Mirlacher, M., Kononen, J., Gasser, T., Mihatsch, M. J., Kallioniemi, O. P. and Sauter, G. (1999). High-Throughput Tissue Microarray Analysis to Evaluate Genes Uncovered by cDNA Microarray Screening in Renal Cell Carcinoma. American Journal of Pathology 154: 981-986.

Monnerat, C., Henriksson, R., Le Chevalier, T., Novello, S., Berthaud, P., Faivre, S. and Raymond, E. (2004). Phase I study of PKC412 (N-benzoyl-staurosporine), a novel oral protein kinase C inhibitor, combined with gemcitabine and cisplatin in patients with non-small-cell lung cancer. Annals of Oncology. 15: 316-323.

Monni, O., Barlund, M., Mousses, S., Kononen, J., Sauter, G., Heiskanen, M., Paavola, P., Avela, K., Chen, Y., Bittner, M. L. and Kallioniemi, A. (2001). Comprehensive copy number and gene expression profiling of the 17q23 amplicon in human breast cancer. Proceedings of the National Academy of Sciences of the United States of America 98: 5711-5716.

Montgomery, E., Argani, P., Hicks, J. L., DeMarzo, A. M. and Meeker, A. K. (2004). Telomere Lengths of Translocation-Associated and Nontranslocation-Associated Sarcomas Differ Dramatically. American Journal of Pathology 164: 1523-1529.

Moro, L., Venturino, M., Bozzo, C., Silengo, L., Altruda, F., Beguinot, L., Tarone, G. and Defilippi, P. (1998). Integrins induce activation of EGF receptor: role in MAP kinase induction and adhesion-dependent cell survival. EMBO Journal 17: 6622-6632.

Moroni, M., Veronese, S., Benvenuti, S., Marrapese, G., Sartore-Bianchi, A., Di Nicolantonio, F., Gambacorta, M., Siena, S. and Bardelli, A. (2005). Gene copy number for epidermal growth factor receptor (EGFR) and clinical response to antiEGFR treatment in colorectal cancer: a cohort study. The Lancet Oncology 6: 279-286.

~~Mosmann, T. (1983). Rapid colorimetric assay for cellular growth and survival: application to proliferation and cytotoxicity assays. Journal of Immunological Methods 65: 55-63.~~

Mousses, S., Bubendorf, L., Wagner, U., Hostetter, G., Kononen, J., Cornelison, R., Goldberger, N., Elkahloun, A. G., Willi, N., Koivisto, P., Ferhle, W., Raffeld, M., Sauter, G. and Kallioniemi, O. P. (2002). Clinical Validation of Candidate Genes Associated with Prostate Cancer Progression in the CWR22 Model System using Tissue Microarrays. Cancer Research 62: 1256-1260.

Mucci, N. R., Akdas, G., Manely, S. and Rubin, M. A. (2000). Neuroendocrine expression in metastatic prostate cancer: evaluation of high throughput tissue microarrays to detect heterogeneous protein expression. Human Pathology 31: 406-414.

~~448~~

~~Bibliography~~

Mukohara, T., Kudoh, S., Yamauchi, S., Kimura, T., Yoshimura, N., Kanazawa, H., Hirata, K., Wanibuchi, H., Fukushima, S., Inoue, K. and Yoshikawa, J. (2003). Expression of epidermal growth factor receptor (EGFR) and downstream-activated peptides in surgically excised non-small-cell lung cancer (NSCLC). Lung Cancer 41: 123-130.

Muller, S., Scaffidi, P., Degryse, B., Bonaldi, T., Ronfani, L., Agresti, A., Beltrame, M. and Bianchi, M. E. (2001). New EMBO members' review: the double life of HMGB1 chromatin protein: architectural factor and extracellular signal. EMBO Journal 20: 4337-4340.

~~Murphree, A. L. and Benedict, W. F. (1984). Retinoblastoma: clues to human oncogenesis. Science 223: 1028-1033.~~

Myhre-Jensen, O., Kaae, S., Madsen, E. H. and Sneppen, O. (1983). Histopathological grading in soft-tissue tumours. Relation to survival in 261 surgically treated patients. Acta Pathologica, Microbiologica, et Immunologica Scandinavica Section A, Pathology 91: 145-150.

Nagayama, S., Katagiri, T., Tsunoda, T., Hosaka, T., Nakashima, Y., Araki, N., Kusuzaki, K., Nakayama, T., Tsuboyama, T., Nakamura, T., Imamura, M., Nakamura, Y. and Toguchida, J. (2002). Genome-wide analysis of gene expression in synovial sarcomas using a cDNA microarray. Cancer Research. 62: 5859-5866.

Nakatani, T., Marui, T., Yamamoto, T., Kurosaka, M., Akisue, T. and Matsumoto, K. (2001). Establishment and characterization of cell line TNMY1 derived from human malignant fibrous histiocytoma. Pathology International. 51: 595-602.

Nestl, A., Von Stein, O. D., Zatloukal, K., Thies, W. G., Herrlich, P., Hofmann, M. and Sleeman, J. P. (2001). Gene expression patterns associated with the metastatic phenotype in rodent and human tumors. Cancer Research 61: 1569-1577.

Ng, T. L., Gown, A. M., Barry, T. S., Cheang, M. C. U., Chan, A. K. W., Turbin, D. A., Hsu, F. D., West, R. B. and Nielsen, T. O. (2004). Nuclear beta-catenin in mesenchymal tumors. Modern Pathology 18: 68-74.

Nielsen, T. O., Hsu, F. D., O'Connell, J. X., Gilks, C. B., Sorensen, P. H., Linn, S., West, R. B., Liu, C. L., Botstein, D., Brown, P. O. and van de Rijn, M. (2003). Tissue microarray validation of epidermal growth factor receptor and SALL2 in synovial sarcoma with comparison to tumors of similar histology. American Journal of Pathology 163: 1449-1456.

Nielsen, T. O., West, R. B., Linn, S. C., Alter, O., Knowling, M. A., O'Connell, J. X., Zhu, S., Fero, M., Sherlock, G. and Pollack, J. R. (2002). Molecular characterisation of soft tissue tumours: a gene expression study. The Lancet 359: 1301-1307.

~~Nilbert, M. and Engellau, J. (2004). Experiences from tissue microarray in soft tissue sarcomas. Acta Orthopaedica Scandinavica. Supplementum. 75: 29-34.~~

~~449~~

~~Bibliography~~

Nilbert, M., Meza-Zepeda, L. A., Francis, P., Berner, J. M., Namlos, H. M., Fernebro, J. and Myklebost, O. (2004). Lessons from genetic profiling in soft tissue sarcomas. Acta Orthopaedica Scandinavica. Supplementum. 75: 35-50.

Nishida, T., Hirota, S., Taniguchi, M., Hashimoto, K., Isozaki, K., Nakamura, H., Kanakura, Y., Tanaka, T., Takabayashi, A., Matsuda, H. and Kitamura, Y. (1998). Familial gastrointestinal stromal tumours with germline mutation of the KIT gene. Nature Genetics 19: 323-324.

Nishio, J., Iwasaki, H., Ishiguro, M., Ohjimi, Y., Fujita, C., Ikegami, H., Ariyoshi, A., Naito, M., Kaneko, Y. and Kikuchi, M. (2003a). Establishment of a novel human dedifferentiated liposarcoma cell line, FU-DDLS-1: conventional and molecular cytogenetic characterization. International Journal of Oncology 22: 535-42.

Nishio, J., Iwasaki, H., Ishiguro, M., Ohjimi, Y., Nishimura, N., Koga, T., Kawarabayashi, T., Kaneko, Y. and Kikuchi, M. (2003b). Establishment of a new human malignant fibrous histiocytoma cell line, FU-MFH-1: cytogenetic characterization by comparative genomic hybridization and fluorescence in situ hybridization. Cancer Genetics & Cytogenetics 144: 44-51.

Noguchi, S. I., Ueki, T., Kawauchi, S., Fukuda, T., Matsuura, H., Sonoda, T. and Tsuneyoshi, M. (1997). Establishment and characterization of a new synovial sarcoma cell line, SN-SY-1: special reference to bcl-2 protein and SYT-SSX1 hybrid transcripts. International Journal of Cancer. 72: 995-1002.

Noseda, M., McLean, G., Niessen, K., Chang, L., Pollet, I., Montpetit, R., Shahidi, R., Dorovini-Zis, K., Li, L., Beckstead, B., Durand, R. E., Hoodless, P. A. and Karsan, A. (2004). Notch Activation Results in Phenotypic and Functional Changes Consistent With Endothelial-to-Mesenchymal Transformation. Circulation Research 94: 910-917.

Nuciforo, P. G., Pellegrini, C., Fasani, R., Maggioni, M., Coggi, G., Parafioriti, A. and Bosari, S. (2003). Molecular and immunohistochemical analysis of HER2/neu oncogene in synovial sarcoma. Human Pathology. 34: 639-645.

O'Sullivan, B., Davis, A. M., Turcotte, R., Bell, R., Catton, C., Chabot, P., Wunder, J., Kandel, R., Goddard, K., Sadura, A., Pater, J. and Zee, B. (2002). Preoperative versus postoperative radiotherapy in soft-tissue sarcoma of the limbs: a randomised trial. Lancet 359: 2235-2241.

Obata, K., Hiraga, H., Nojima, T., Yoshida, M. C. and Abe, S. (1999). Molecular characterization of the genomic breakpoint junction in a t(11;22) translocation in Ewing sarcoma. Genes, Chromosomes & Cancer 25: 6-15.

Oberthuer, A., Berthold, F., Warnat, P., Hero, B., Kahlert, Y., Spitz, R., Ernestus, K., Konig, R., Haas, S., Eils, R., Schwab, M., Brors, B., Westermann, F. and Fischer, M. (2006). Customized Oligonucleotide Microarray Gene Expression-Based Classification of Neuroblastoma Patients Outperforms Current Clinical Risk Stratification. Journal of Clinical Oncology 24: 5070-5078.

~~450~~

~~Bibliography~~

Oda, Y., Saito, T., Tateishi, N., Ohishi, Y., Tamiya, S., Yamamoto, H., Yokoyama, R., Uchiumi, T., Iwamoto, Y., Kuwano, M. and Tsuneyoshi, M. (2005). ATP-binding cassette superfamily transporter gene expression in human soft tissue sarcomas. International Journal of Cancer 114: 854-862.

~~Oh, S. P. and Li, E. (1997). The signaling pathway mediated by the type IIB activin receptor controls axial patterning and lateral asymmetry in the mouse. Genes and Development 11: 1812-1826.~~

Okabe, H., Satoh, S., Kato, T., Kitahara, O., Yanagawa, R., Yamaoka, Y., Tsunoda, T., Furukawa, Y. and Nakamura, Y. (2001). Genome-wide analysis of gene expression in human hepatocellular carcinomas using cDNA microarray: identification of genes involved in viral carcinogenesis and tumor progression. Cancer Research. 61: 2129-37.

~~Okuno, S. (2006). Mammalian target of rapamycin inhibitors in sarcomas. Current Opinion in Oncology 18: 360-362.~~

~~Oliveira, A. M., Okuno, S. H., Nascimento, A. G. and Lloyd, R. V. (2003). Skp2 protein expression in soft tissue sarcomas. Journal of Clinical Oncology. 21: 722-727.~~

~~Olivier, M., Eeles, R., Hollstein, M., Khan, M. A., Harris, C. C. and Hainaut, P. (2002). The IARC TP53 database: new online mutation analysis and recommendations to users. Human Mutation 19: 607-614.~~

Osman, I., Bajorin, D. F., Sun, T.-T., Zhong, H., Douglas, D., Scattergood, J., Zheng, R., Han, M., Marshall, K. W. and Liew, C.-C. (2006). Novel Blood Biomarkers of Human Urinary Bladder Cancer. Clinical Cancer Research 12: 3374-3380.

Otano-Joos, M., Mechtersheimer, G., Ohl, S., Wilgenbus, K. K., Scheurlen, W., Lehnert, T., Willeke, F., Otto, H. F., Lichter, P. and Joos, S. (2000). Detection of chromosomal imbalances in leiomyosarcoma by comparative genomic hybridization and interphase cytogenetics. Cytogenetics & Cell Genetics 90: 86-92.

Ow, K., Delprado, W., Fisher, R., Barrett, J., Yu, Y., Jackson, P. and Russell, P. J. (2000). Relationship between expression of the KAI1 metastasis suppressor and other markers of advanced bladder cancer. Journal of Pathology 191: 39-47.

~~Packeisen, J., Korsching, E., Herbst, H., Boecker, W. and Buerger, H. (2003). Demystified ... Tissue microarray technology. Molecular Pathology. 56: 198-204.~~

Paez, J. G., Janne, P. A., Lee, J. C., Tracy, S., Greulich, H., Gabriel, S., Herman, P., Kaye, F. J., Lindeman, N., Boggon, T. J., Naoki, K., Sasaki, H., Fujii, Y., Eck, M. J., Sellers, W. R., Johnson, B. E. and Meyerson, M. (2004). EGFR mutations in lung cancer: correlation with clinical response to gefitinib therapy. Science 304: 1497-1500.

Pai, R., Soreghan, B., Szabo, I. L., Pavelka, M., Baatar, D. and Tarnawski, A. S. (2002). Prostaglandin E2 transactivates EGF receptor: a novel mechanism for promoting colon cancer growth and gastrointestinal hypertrophy. Nature Medicine 8: 289-293.

~~451~~

~~Bibliography~~

Parente, F., Grosgeorge, J., Coindre, J. M., Terrier, P., Vilain, O. and Turc-Carel, C. (1999). Comparative genomic hybridization reveals novel chromosome deletions in 90 primary soft tissue tumors. Cancer Genetics & Cytogenetics. 115: 89-95.

Park, H., Adams, M. A., Lachat, P., Bosman, F., Pang, S. C. and Graham, C. H. (2000). Hypoxia induces the expression of a 43-kDa protein (PROXY-1) in normal and malignant cells. Biochemical & Biophysical Research Communications 276: 321-328.

~~Pearson, A. D. J., Craft, A. W., Ratcliffe, J. M., Birch, J. M., Morris Jones, P. H. and Roberts, D. F. (1982). Two families with the Li-Fraumeni cancer family syndrome. J Med Genet 19: 362-365.~~

Penzel, R., Aulmann, S., Moock, M., Schwarzbach, M., Rieker, R. J. and Mechtersheimer, G. (2005). The location of KIT and PDGFRA gene mutations in gastrointestinal stromal tumours is site and phenotype associated. Journal of Clinical Pathology 58: 634-639.

Perez-Losada, J., Pintado, B., Gutierrez-Adan, A., Flores, T., Banares-Gonzalez, B., del Campo, J. C., Martin-Martin, J. F., Battaner, E. and Sanchez-Garcia, I. (2000). The chimeric FUS/TLS-CHOP fusion protein specifically induces liposarcomas in transgenic mice. Oncogene 19: 2413-2422.

Perez-Soler, R., Chachoua, A., Hammond, L. A., Rowinsky, E. K., Huberman, M., Karp, D., Rigas, J., Clark, G. M., Santabarbara, P. and Bonomi, P. (2004). Determinants of tumor response and survival with erlotinib in patients with non--small-cell lung cancer. Journal of Clinical Oncology 22: 3238-3247.

Perou, C. M., Sorlie, T., Eisen, M. B., van de Rijn, M., Jeffrey, S. S., Rees, C. A., Pollack, J. R., Ross, D. T., Johnsen, H., Akslen, L. A., Fluge, O., Pergamenschikov, A., Williams, C., Zhu, S. X., Lonning, P. E., Borresen-Dale, A. L., Brown, P. O. and Botstein, D. (2000). Molecular portraits of human breast tumours. Nature. 406: 747- 752.

Perrem, K., Bryan, T. M., Englezou, A., Hackl, T., Moy, E. L. and Reddel, R. R. (1999). Repression of an alternative mechanism for lengthening of telomeres in somatic cell hybrids. Oncogene 18: 3383-90.

Perrem, K., Colgin, L. M., Neumann, A. A., Yeager, T. R. and Reddel, R. R. (2001). Coexistence of alternative lengthening of telomeres and telomerase in hTERT- transfected GM847 cells. Molecular & Cellular Biology 21: 3862-75.

~~Pfaffl, M. W. (2001). A new mathematical model for relative quantification in real-time RT-PCR. Nucleic Acids Research. 29: 2002-2007.~~

Pfaffl, M. W., Horgan, G. W. and Dempfle, L. (2002). Relative expression software tool (REST(C)) for group-wise comparison and statistical analysis of relative expression results in real-time PCR. Nucleic Acids Research. 30: e36-.

~~452~~

~~Bibliography~~

~~Pisters, P. W. (2002). Staging and Prognosis. American Cancer Society Atlas of Clinical Oncology: Soft tissue Sarcomas. R. E. Pollock. Hamilton, Ontario, BC Decker Inc: 80- 88.~~

Pisters, P. W., Leung, D. H., Woodruff, J., Shi, W. and Brennan, M. F. (1996). Analysis of prognostic factors in 1,041 patients with localized soft tissue sarcomas of the extremities. Journal of Clinical Oncology 14: 1679-89.

~~Pisters, P. W. T., Ballo, M. T. and Patel, S. R. (2002). Preoperative Chemoradiation Treatment Strategies for Localized Sarcoma. Annals of Surgical Oncology. 9: 535-542.~~

Pitti, R. M., Marsters, S. A., Lawrence, D. A., Roy, M., Kischkel, F. C., Dowd, P., Huang, A., Donahue, C. J., Sherwood, S. W., Baldwin, D. T., Godowski, P. J., Wood, W. I., Gurney, A. L., Hillan, K. J., Cohen, R. L., Goddard, A. D., Botstein, D. and Ashkenazi, A. (1998). Genomic amplification of a decoy receptor for Fas ligand in lung and colon cancer. Nature 396: 699-703.

~~Plumb, J. A. (2004). Cell Sensitivity Assays: The MTT Assay. Cancer Cell Culture: 165-169.~~

~~Politi, K., Feirt, N. and Kitajewski, J. (2004). Notch in mammary gland development and breast cancer. Seminars in Cancer Biology. 14: 341-347.~~

~~Pollock, R. (2001). Soft Tissue Saroma: Setting the Stage. Annals of Surgical Oncology. 8: 273-274.~~

~~Pollock, R. E. (2002). American Cancer Society Atlas of Clinical Oncology: Soft Tissue Sarcomas. London, BC Decker Inc.~~

~~Pollock, R. E., Lang, A., Luo, J., El-Naggar, A. K. and Yu, D. (1996). Soft tissue sarcoma metastasis from clonal expansion of p53 mutated tumor cells. Oncogene 12: 2035-2039.~~

~~Portwine, C., Lees, J., Verselis, S., Li, F. P. and Malkin, D. (2000). Absence of germline p16(INK4a) alterations in p53 wild type Li-Fraumeni syndrome families. Journal of Medical Genetics 37.~~

Potti, A., Ganti, A. K., Tendulkar, K., Sholes, K., Chitajallu, S., Koch, M. and Kargas, S. (2004). Determination of vascular endothelial growth factor (VEGF) overexpression in soft tissue sarcomas and the role of overexpression in leiomyosarcoma. Journal of Cancer Research & Clinical Oncology. 130: 52-6.

Prasad, N. B., Biankin, A. V., Fukushima, N., Maitra, A., Dhara, S., Elkahloun, A. G., Hruban, R. H., Goggins, M. and Leach, S. D. (2005). Gene Expression Profiles in Pancreatic Intraepithelial Neoplasia Reflect the Effects of Hedgehog Signaling on Pancreatic Ductal Epithelial Cells. Cancer Research 65: 1619-1626.

~~Prenen, H., Cools, J., Mentens, N., Folens, C., Sciot, R., Schoffski, P., Van Oosterom, A., Marynen, P. and Debiec-Rychter, M. (2006). Efficacy of the kinase inhibitor~~

~~453~~

~~Bibliography~~

~~SU11248 against gastrointestinal stromal tumor mutants refractory to imatinib mesylate. Clinical Cancer Research 12: 2622-2627.~~

Pretto, D., Barco, R., Rivera, J., Neel, N., Gustavson, M. D. and Eid, J. E. (2006). The synovial sarcoma translocation protein SYT-SSX2 recruits [beta]-catenin to the nucleus and associates with it in an active complex. Oncogene 25: 3661-3669.

Psyrri, A., Kassar, M., Yu, Z., Bamias, A., Weinberger, P. M., Markakis, S., Kowalski, D., Camp, R. L., Rimm, D. L. and Dimopoulos, M. A. (2005). Effect of epidermal growth factor receptor expression level on survival in patients with epithelial ovarian cancer. Clinical Cancer Research 11: 8637-8643.

~~Putnam Jnr, J. B. (2002). Metastases from Sarcoma. Soft Tissue Sarcomas. R. E. Pollock. Hamilton, Ontario, BC Decker Inc: 262-278.~~

~~Quackenbush, J. (2002). Microarray data normalization and transformation. Nature Genetics. 32: 496-501.~~

Radonic, A., Thulke, S., Mackay, I. M., Landt, O., Siegert, W. and Nitsche, A. (2004). Guideline to reference gene selection for quantitative real-time PCR. Biochemical & Biophysical Research Communications 313: 856-862.

~~Radtke, F. and Raj, K. (2003). The role of Notch in tumorigenesis: oncogene or tumour suppressor? Nature Reviews Cancer 3: 756-767.~~

Rainen, L., Oelmueller, U., Jurgensen, S., Wyrich, R., Ballas, C., Schram, J., Herdman, C., Bankaitis-Davis, D., Nicholls, N., Trollinger, D. and Tryon, V. (2002). Stabilization of mRNA Expression in Whole Blood Samples. Clinical Chemistry 48: 1883-1890.

~~Ramaswamy, S., Ross, K. N., Lander, E. S. and Golub, T. R. (2003). A molecular signature of metastasis in primary solid tumors. Nature Genetics 33: 49-54.~~

Ramaswamy, S., Tamayo, P., Rifkin, R., Mukherjee, S., Yeang, C. H., Angelo, M., Ladd, C., Reich, M., Latulippe, E., Mesirov, J. P., Poggio, T., Gerald, W., Loda, M., Lander, E. S. and Golub, T. R. (2001). Multiclass cancer diagnosis using tumor gene expression signatures. Proceedings of the National Academy of Sciences of the United States of America 98: 15149-15154.

~~Rasheed, S., Nelson-Rees, W. A., Toth, E. M., Arnstein, P. and Gardner, M. B. (1974). Characterization of a newly derived human sarcoma cell line (HT-1080). Cancer. 33: 1027-1033.~~

~~Reddel, R. R. (2000). The role of senescence and immortalization in carcinogenesis. Carcinogenesis 21: 477-484.~~

~~Reddel, R. R. (2003). Alternative lengthening of telomeres, telomerase, and cancer. Cancer Letters 194: 155-62.~~

~~454~~

~~Bibliography~~

~~Reddy, M. P., Reddy, P. and Lilien, D. L. (2003). F-18 FDG PET Imaging in Gastrointestinal Stromal Tumor. Clinical Nuclear Medicine. 28: 677-679.~~

~~Ren, B., Yu, Y. P., Jing, L., Liu, L., Michalopoulos, G. K., Luo, J. H. and Rao, U. N. (2003). Gene expression analysis of human soft tissue leiomyosarcomas. Human Pathology 34: 549-558.~~

~~Rhim, J. S. (2000). Development of Human Cell Lines from Multiple Organs. Annals of the New York Academy of Sciences 919: 16-25.~~

Rhodes, D. R., Barrette, T. R., Rubin, M. A., Ghosh, D. and Chinnaiyan, A. M. (2002). Meta-analysis of microarrays: interstudy validation of gene expression profiles reveals pathway dysregulation in prostate cancer. Cancer Research 62: 4427-4433.

Ricci, C., Landuzzi, L., Rossi, I., De Giovanni, C., Nicoletti, G., Astolfi, A., Pupa, S., Menard, S., Scotlandi, K., Nanni, P. and Lollini, P.-L. (2000). Expression of HER/erbb family of receptor tyrosine kinases and induction of differentiation by glial growth factor 2 in human rhabdomyosarcoma cells. International Journal of Cancer 87: 29-36.

Richter, J., Wagner, U., Kononen, J., Fijan, A., Bruderer, J., Schmid, U., Ackermann, D., Maurer, R., Alund, G., Knonagel, H., Rist, M., Wilber, K., Anabitarte, M., Hering, F., Hardmeier, T., Schonenberger, A., Flury, R., Jager, P., Luc Fehr, J., Schraml, P., Moch, H., Mihatsch, M. J., Gasser, T., Kallioniemi, O. P. and Sauter, G. (2000). High- Throughput Tissue Microarray Analysis of Cyclin E Gene Amplification and Overexpression in Urinary Bladder Cancer. American Journal of Pathology 157: 787- 794.

Rieske, P., Bartkowiak, J. K., Szadowska, A. M., Olborski, B., Harezga-Bal, B. and Debiec-Rychter, M. (1999). A comparative study of P53/MDM2 genes alterations and P53/MDM2 proteins immunoreactivity in soft-tissue sarcomas. Journal of Experimental & Clinical Cancer Research. 18: 403-416.

~~Rimm, D. L., Camp, R. L., Charette, L. A., Costa, J., Olsen, D. A. and Reiss, M. (2001). Tissue microarray: a new technology for amplification of tissue resources. Cancer Journal 7: 24-31.~~

Risbridger, G. P., Ball, E. M., Wang, H., Mellor, S. L. and Peehl, D. M. (2004). Re- evaluation of inhibin alpha subunit as a tumour suppressor in prostate cancer. Molecular and Cellular Endocrinology 225: 73-76.

Roberts, A. B. and Wakefield, L. M. (2003). The two faces of transforming growth factor beta in carcinogenesis. Proceedings of the National Academy of Sciences of the United States of America 100: 8621-8623.

~~Rodan, G. A. and Harada, S. (1997). The missing bone. Cell 89: 677-680.~~

Rogan, E. M., Bryan, T. M., Hukku, B., Maclean, K., Chang, A.-M., Moy, E. L., Englezou, A., Warneford, S. G., Dalla-Pozza, L. and Reddel, R. R. (1995). Alterations in p53 and p16(INK4) expression and telomere length during spontaneous

~~455~~

~~Bibliography~~

~~immortalization of Li-Fraumeni syndrome fibroblasts. Molecular & Cellular Biology 15: 4745-4753.~~

Romond, E. H., Perez, E. A., Bryant, J., Suman, V. J., Geyer, C. E. J., Davidson, N. E., Tan-Chiu, E., Martino, S., Paik, S., Kaufman, P. A., Swain, S. M., Pisansky, T. M., Fehrenbacher, L., Kutteh, L. A., Vogel, V. G., Visscher, D. W., Yothers, G., Jenkins, R. B., Brown, A. M., Dakhil, S. R., Mamounas, E. P., Lingle, W. L., Klein, P. M., Ingle, J. N. and Wolmark, N. (2005). Trastuzumab plus adjuvant chemotherapy for operable HER2-positive breast cancer. The New England Journal of Medicine. 353: 1673-1684.

Rosenwald, A., Wright, G., Chan, W. C., Connors, J. M., Campo, E., Fisher, R. I., Gascoyne, R. D., Muller-Hermelink, H. K., Smeland, E. B., Giltnane, J. M., Hurt, E. M., Zhao, H., Averett, L., Yang, L., Wilson, W. H., Jaffe, E. S., Simon, R., Klausner, R. D., Powell, J., Duffey, P. L., Longo, D. L., Greiner, T. C., Weisenburger, D. D., Sanger, W. G., Dave, B. J., Lynch, J. C., Vose, J., Armitage, J. O., Montserrat, E., Lopez- Guillermo, A., Grogan, T. M., Miller, T. P., LeBlanc, M., Ott, G., Kvaloy, S., Delabie, J., Holte, H., Krajci, P., Stokke, T. and Staudt, L. M. (2002). The use of molecular profiling to predict survival after chemotherapy for diffuse large-B-cell lymphoma. New England Journal of Medicine 346: 1937-1947.

Ross, D. T., Scherf, U., Eisen, M. B., Perou, C. M., Rees, C., Spellman, P., Iyer, V., Jeffrey, S. S., Van de Rijn, M., Waltham, M., Pergamenschikov, A., Lee, J. C., Lashkari, D., Shalon, D., Myers, T. G., Weinstein, J. N., Botstein, D. and Brown, P. O. (2000). Systematic variation in gene expression patterns in human cancer cell lines. Nature Genetics. 24: 227-235.

Rossi, C. R., Mocellin, S., Pilati, P., Foletto, M., Nitti, D. and Lise, M. (2003). TNFalpha-based isolated perfusion for limb-threatening soft tissue sarcomas: state of the art and future trends. Journal of Immunotherapy. 26: 291-300.

Rubin, B. P., Singer, S., Tsao, C., Duensing, A., Lux, M. L., Ruiz, R., Hibbard, M. K., Chen, C.-J., Xiao, S., Tuveson, D. A., Demetri, G. D., Fletcher, C. D. M. and Fletcher, J. A. (2001). KIT Activation Is a Ubiquitous Feature of Gastrointestinal Stromal Tumors. Cancer Research 61: 8118-8121.

~~Rubin, M. A., Dunn, R., Strawderman, M. and Pienta, K. J. (2002). Tissue Microarray Sampling Strategy for Prostate Cancer Biomarker Analysis. American Journal of Surgical Pathology 26: 312-319.~~

~~Ryan, A. J. and Wedge, S. R. (2005). ZD6474--a novel inhibitor of VEGFR and EGFR tyrosine kinase activity. British Journal of Cancer 92: S6-13.~~

~~Sabah, M., Cummins, R., Leader, M. and Kay, E. (2006). Aberrant expression of the Rb pathway proteins in soft tissue sarcomas. Applied Immunohistochemistry & Molecular Morphology 14: 397-403.~~

~~Saito, T., Oda, Y., Kawaguchi, K. I., Sugimachi, K., Yamamoto, H., Tateishi, N., Tanaka, K., Matsuda, S., Iwamoto, Y., Ladanyi, M. and Tsuneyoshi, M. (2004). E-~~

~~456~~

~~Bibliography~~

~~cadherin mutation and Snail overexpression as alternative mechanisms of E-cadherin inactivation in synovial sarcoma. Oncogene 23: 8629-8638.~~

Saito, T., Oda, Y., Sakamoto, A., Tamiya, S., Kinukawa, N., Hayashi, K., Iwamoto, Y. and Tsuneyoshi, M. (2000). Prognostic value of the preserved expression of the E- cadherin and catenin families of adhesion molecules and of beta-catenin mutations in synovial sarcoma. Journal of Pathology 192: 342-350.

Saito, T., Oda, Y., Tanaka, K., Matsuda, S., Tamiya, S., Iwamoto, Y. and Tsuneyoshi, M. (2001). beta-catenin nuclear expression correlates with cyclin D1 overexpression in sporadic desmoid tumours. Journal of Pathology 195: 222-228.

Saito, T., Oda, Y., Yamamoto, H., Kawaguchi, K., Tanaka, K., Matsuda, S., Iwamoto, Y. and Tsuneyoshi, M. (2006). Nuclear beta-catenin correlates with cyclin D1 expression in spindle and pleomorphic sarcomas but not in synovial sarcoma. Human Pathology 37: 689-697.

Sakabe, T., Shinomiya, T., Mori, T., Ariyama, Y., Fukuda, Y., Fujiwara, T., Nakamura, Y. and Inazawa, J. (1999). Identification of a novel gene, MASL1, within an amplicon at 8p23.1 detected in malignant fibrous histiocytomas by comparative genomic hybridization. Cancer Research 59: 511-515.

Sakakura, C., Hagiwara, A., Nakanishi, M., Shimomura, K., Takagi, T., Yasuoka, R., Fujita, Y., Abe, T., Ichikawa, Y., Takahashi, S., Ishikawa, T., Nishizuka, I., Morita, T., Shimada, H., Okazaki, Y., Hayashizaki, Y. and Yamagishi, H. (2002). Differential gene expression profiles of gastric cancer cells established from primary tumour and malignant ascites. British Journal of Cancer. 87: 1153-61.

Sakurai, S., Fukayama, M., Kaizaki, Y., Saito, K., Kanazawa, K., Kitamura, M., Iwasaki, Y., Hishima, T., Hayashi, Y. and Koike, M. (1998). Telomerase activity in gastrointestinal stromal tumors. Cancer 83: 2060-2066.

Sallinen, S.-L., Sallinen, P. K., Haapasalo, H. K., Helin, H. J., Helen, P. T., Schraml, P., Kallioniemi, O.-P. and Kononen, J. (2000). Identification of Differentially Expressed Genes in Human Gliomas by DNA Microarray and Tissue Chip Techniques. Cancer Research 60: 6617-6622.

Salnikow, K., Costa, M., Figg, W. D. and Blagosklonny, M. V. (2000). Hyperinducibility of Hypoxia-responsive Genes without p53/p21-dependent Checkpoint in Aggressive Prostate Cancer. Cancer Research 60: 5630-5634.

Sanchez-Carbayo, M., Socci, N. D., Charytonowicz, E., Lu, M., Prystowsky, M., Childs, G. and Cordon-Cardo, C. (2002). Molecular profiling of bladder cancer using cDNA microarrays: defining histogenesis and biological phenotypes. Cancer Research 62: 6973-6980.

~~Sanchez-Carbayo, M., Socci, N. D., Lozano, J. J., Li, W., Charytonowicz, E., Belbin, T. J., Prystowsky, M. B., Ortiz, A. R., Childs, G. and Cordon-Cardo, C. (2003). Gene~~

~~457~~

~~Bibliography~~

~~discovery in bladder cancer progression using cDNA microarrays. American Journal of Pathology. 163: 505-516.~~

Sato, O., Wada, T., Kawai, A., Yamaguchi, U., Makimoto, A., Kokai, Y., Yamashita, T., Chuman, H., Beppu, Y., Tani, Y. and Hasegawa, T. (2005). Expression of epidermal growth factor receptor, ERBB2 and KIT in adult soft tissue sarcomas: a clinicopathologic study of 281 cases. Cancer 103: 1881-1890.

~~Scaltriti, M. and Baselga, J. (2006). The Epidermal Growth Factor Receptor Pathway: A Model for Targeted Therapy. Clinical Cancer Research 12: 5268-5272.~~

~~Scappaticci, F. A. (2002). Mechanisms and future directions for angiogenesis-based cancer therapies. Journal of Clinical Oncology 20: 3906-3927.~~

~~Scappaticci, F. A. and Marina, N. (2001). New molecular targets and biological therapies in sarcomas. Cancer Treatment Reviews 27: 317-26.~~

Schaefer, K. L., Brachwitz, K., Braun, Y., Diallo, R., Wai, D. H., Zahn, S., Schneider, D. T., Kuhnen, C., Vollmann, A., Brockhoff, G., Gabbert, H. E. and Poremba, C. (2006). Constitutive activation of neuregulin/ERBB3 signaling pathway in clear cell sarcoma of soft tissue. Neoplasia 8: 613-622.

Schaefer, K. L., Brachwitz, K., Wai, D. H., Braun, Y., Diallo, R., Korsching, E., Eisenacher, M., Voss, R., Van Valen, F., Baer, C., Selle, B., Spahn, L., Liao, S. K., Lee, K. A., Hogendoorn, P. C., Reifenberger, G., Gabbert, H. E. and Poremba, C. (2004). Expression profiling of t(12;22) positive clear cell sarcoma of soft tissue cell lines reveals characteristic up-regulation of potential new marker genes including ERBB3. Cancer Research 64: 3395-3405.

Schaefer, K. L., Wai, D. H., Poremba, C., Korsching, E., van Valen, F., Ozaki, T., Boecker, W. and Dockhorn-Dworniczak, B. (2002). Characterization of the malignant melanoma of soft-parts cell line GG-62 by expression analysis using DNA microarrays. Virchows Archives 440: 476-484.

Schlueter, C., Weber, H., Meyer, B., Rogalla, P., Roser, K., Hauke, S. and Bullerdiek, J. (2005). Angiogenetic signaling through hypoxia: HMGB1: an angiogenetic switch molecule. American Journal of Pathology 166: 1259-1263.

Schmidt, H., Taubert, H., Wurl, P., Kappler, M., Lange, H., Bartel, F., Bache, M., Holzhausen, H. J. and Hinze, R. (2002). Gains of 12q are the most frequent genomic imbalances in adult fibrosarcoma and are correlated with a poor outcome. Genes, Chromosomes & Cancer 34: 69-77.

Schmidt, H., Wurl, P., Taubert, H., Meye, A., Bache, M., Holzhausen, H. J. and Hinze, R. (1999). Genomic imbalances of 7p and 17q in malignant peripheral nerve sheath tumors are clinically relevant. Genes, Chromosomes & Cancer 25: 205-211.

~~Schmookler, B., Bickels, J., Jellinek, J., Sugarbaker, P. and Malawer, M. (2001). Bone and soft tissue sarcomas: epidemiology, radiology, pathology and fundamentals of~~

~~458~~

~~Bibliography~~

~~surgical treatment. Musculoskeletal Cancer Surgery. Dordrecht, Kluwer Academic Publishers: 3-37.~~

~~Schnabel, M., Fichtel, I., Gotzen, L. and Schlegel, J. (2002). Differential expression of Notch genes in human osteoblastic cells. International Journal of Molecular Medicine 9: 229-232.~~

Schneider-Stock, R., Boltze, C., Lasota, J., Miettinen, M., Peters, B., Pross, M., Roessner, A. and Gunther, T. (2003). High prognostic value of p16INK4 alterations in gastrointestinal stromal tumors. Journal of Clinical Oncology 21: 1688-1697.

Schraml, P., Kononen, J., Bubendorf, L., Moch, H., Bissig, H., Nocito, A., Mihatsch, M. J., Kallioniemi, O.-P. and Sauter, G. (1999). Tissue Microarrays for Gene Amplification Surveys in Many Different Tumor Types. Clinical Cancer Research 5: 1966-1975.

~~Schuyer, M. and Berns, E. M. (1999). Is TP53 dysfunction required for BRCA1- associated carcinogenesis? Molecular Cell Endocrinology 155: 143-152.~~

~~Scoggins, C. R. and Pollock, R. E. (2005). Extremity soft tissue sarcoma: evidence- based multidisciplinary management. Journal of Surgical Oncology 90: 10-13.~~

~~Scully, R. and Livingston, D. M. (2000). In search of the tumour-suppressor functions of BRCA1 and BRCA2. Nature 408: 429-432.~~

Segal, N. H., Pavlidis, P., Antonescu, C. R., Maki, R. G., Noble, W. S., DeSantis, D., Woodruff, J. M., Lewis, J. J., Brennan, M. F., Houghton, A. N. and Cordon-Cardo, C. (2003). Classification and Subtype Prediction of Adult Soft Tissue Sarcoma by Functional Genomics. American Journal of Pathology 163: 691-700.

Sharma, S. V., Bell, D. W., Settleman, J., Haber, D. A., Sharma, S. V., Bell, D. W., Settleman, J. and Haber, D. A. (2007). Epidermal growth factor receptor mutations in lung cancer. Nature Reviews Cancer 7: 169-81.

Sharpless, N. E., Ferguson, D. O., O'Hagan, R. C., Castrillon, D. H., Lee, C., Farazi, P. A., Alson, S., Fleming, J., Morton, C. C., Frank, K., Chin, L., Alt, F. W. and DePinho, R. A. (2001). Impaired nonhomologous end-joining provokes soft tissue sarcomas harboring chromosomal translocations, amplifications, and deletions. Molecular Cell 8: 1187-1196.

Shaw, L., Chao, C., Wewer, U. and Mercurio, A. (1996). Function of the integrin alpha 6 beta 1 in metastatic breast carcinoma cells assessed by expression of a dominant- negative receptor. Cancer Research 56: 959-963.

~~Shay, J. W. and Bacchetti, S. (1997). A survey of telomerase activity in human cancer. European Journal of Cancer 33: 787-91.~~

~~Shen, M. H., Harper, P. S. and Upadhyaya, M. (1996). Molecular genetics of neurofibromatosis type 1 (NF1). Journal of Medical Genetics 33: 2-17.~~

~~459~~

~~Bibliography~~

Shepherd, F. A., Rodrigues Pereira, J., Ciuleanu, T., Tan, E. H., Hirsh, V., Thongprasert, S., Campos, D., Maoleekoonpiroj, S., Smylie, M., Martins, R., van Kooten, M., Dediu, M., Findlay, B., Tu, D., Johnston, D., Bezjak, A., Clark, G., Santabarbara, P., Seymour, L. and the National Cancer Institute of Canada Clinical Trials, G. (2005). Erlotinib in Previously Treated Non-Small-Cell Lung Cancer. New England Journal of Medicine 353: 123-132.

~~Sherlock, G. (2005). Of fish and chips. Nature Methods 2: 329-330.~~

~~Shi, W. and Harris, A. (2006). Notch Signaling in Breast Cancer and Tumor Angiogenesis: Cross-Talk and Therapeutic Potentials. Journal of Mammary Gland Biology and Neoplasia 11: 41-52.~~

Shin, B. K., Wang, H., Yim, A. M., Le Naour, F., Brichory, F., Jang, J. H., Zhao, R., Puravs, E., Tra, J., Michael, C. W., Misek, D. E. and Hanash, S. M. (2003). Global Profiling of the Cell Surface Proteome of Cancer Cells Uncovers an Abundance of Proteins with Chaperone Function. Journal of Biological Chemistry 278: 7607-7616.

Shin, H. J., Kalapurakal, S. K., Lee, J. J., Ro, J. Y., Hong, W. K. and Lee, J. S. (1997). Comparison of p53 immunoreactivity in fresh-cut versus stored slides with and without microwave heating. Modern Pathology 10: 224-230.

~~Shiu, M. H. and Brennan, M. F. (1989). Biopsy of Soft Tissue Sarcoma. Surgical Management of Soft Tissue Sarcoma. M. H. Shiu and M. F. Brennan. Philadelphia, Lea & Febiger: 58-63.~~

Shmulevich, I., Hunt, K., El-Naggar, A., Taylor, E., Ramdas, L., Laborde, P., Hess, K. R., Pollock, R. and Zhang, W. (2002). Tumor specific gene expression profiles in human leiomyosarcoma: an evaluation of intratumor heterogeneity. Cancer. 94: 2069- 75.

Sibilia, M., Kroismayr, R., Lichtenberger, B. M., Natarajan, A., Hecking, M. and Holcmann, M. (2007). The epidermal growth factor receptor: from development to tumorigenesis. Differentiation 75: 770-787.

Siegel-Lakhai, W. S., Beijnen, J. H. and Schellens, J. H. M. (2005). Current Knowledge and Future Directions of the Selective Epidermal Growth Factor Receptor Inhibitors Erlotinib (Tarceva(R)) and Gefitinib (Iressa(R)). The Oncologist 10: 579-589.

~~Sim, F. H., Edmonson, J. H. and Wold, L. E. (1993). Soft-tissue sarcomas. Future perspectives. Clinical Orthopaedics & Related Research 289: 106-112.~~

~~Singer, S., Demetri, G. D., Baldini, E. H. and Fletcher, C. D. (2000). Management of soft-tissue sarcomas: an overview and update. Lancet Oncology 1: 75-85.~~

Singh, D., Febbo, P. G., Ross, K., Jackson, D. G., Manola, J., Ladd, C., Tamayo, P., Renshaw, A. A., D'Amico, A. V., Richie, J. P., Lander, E. S., Loda, M., Kantoff, P. W., Golub, T. R. and Sellers, W. R. (2002). Gene expression correlates of clinical prostate cancer behavior. Cancer Cell 1: 203-209.

~~460~~

~~Bibliography~~

~~Sirover, M. A. (1999). New insights into an old protein: the functional diversity of mammalian glyceraldehyde-3-phosphate dehydrogenase. Biochimica et Biophysica Acta 2: 159-84.~~

Skacel, M., Skilton, B., Pettay, J. D. and Tubbs, R. R. (2002). Tissue microarrays: a powerful tool for high-throughput analysis of clinical specimens: a review of the method with validation data. Applied Immunohistochemistry & Molecular Morphology 10: 1-6.

~~Skapek, S. X. and Chui, C. H. (2000). Cytogenetics and the biologic basis of sarcomas. Current Opinion in Oncology 12: 315-22.~~

~~Slavin, J. L. and Thomas, D. (2006). Gene expression profiling of sarcomas. Pathology 38: 101-102.~~

Sloan, K. E., Eustace, B. K., Stewart, J. K., Zehetmeier, C., Torella, C., Simeone, M., Roy, J. E., Unger, C., Louis, D. N., Ilag, L. L. and Jay, D. G. (2004). CD155/PVR plays a key role in cell motility during tumor cell invasion and migration. BMC Cancer 4.

~~Slonim, D. K. (2002). From patterns to pathways: gene expression data analysis comes of age. Nature Genetics. 32: 502-508.~~

Slovak, M. L., Coccia, M., Meltzer, P. S. and Trent, J. M. (1991). Molecular analysis of two human doxorubicin-resistant cell lines: evidence for differing multidrug resistance mechanisms. Anticancer Research. 11: 423-8.

Slovak, M. L., Ho, J. P., Bhardwaj, G., Kurz, E. U., Deeley, R. G. and Cole, S. P. (1993). Localization of a novel multidrug resistance-associated gene in the HT1080/DR4 and H69AR human tumor cell lines. Cancer Research. 53: 3221-5.

~~Smyth, G. K. (2004). Linear models and empirical Bayes methods for assessing differential expression in microarray experiments. Statistical Applications in Genetics and Molecular Biology. 3.~~

~~Smyth, G. K. and Speed, T. (2003). Normalization of cDNA microarray data. Methods 31: 265-273.~~

Solit, D. B., Garraway, L. A., Pratilas, C. A., Sawai, A., Getz, G., Basso, A., Ye, Q., Lobo, J. M., She, Y., Osman, I., Golub, T. R., Sebolt-Leopold, J., Sellers, W. R. and Rosen, N. (2006). BRAF mutation predicts sensitivity to MEK inhibition. Nature 439: 358-362.

Sonobe, H., Takeuchi, T., Liag, S. B., Taguchi, T., Yuri, K., Shimizu, K., Iwata, J., Furihata, M., Ohtsuki, Y. and Testa, J. R. (1999). A new human synovial sarcoma cell line, HS-SY-3, with a truncated form of hybrid SYT/SSX1 gene. International Journal of Cancer. 82: 459-64.

~~461~~

~~Bibliography~~

Sorensen, B. S., Torring, N., Bor, M. V. and Nexo, E. (2004). The DNA damaging agent VP16 induces the expression of a subset of ligands from the EGF system in bladder cancer cells, whereas none of the four EGF receptors are induced. Molecular and Cellular Biochemistry 260: 129-135.

Sorlie, T., Perou, C. M., Tibshirani, R., Aas, T., Geisler, S., Johnsen, H., Hastie, T., Eisen, M. B., van de Rijn, M., Jeffrey, S. S., Thorsen, T., Quist, H., Matese, J. C., Brown, P. O., Botstein, D., Eystein Lonning, P. and Borresen-Dale, A. L. (2001). Gene expression patterns of breast carcinomas distinguish tumor subclasses with clinical implications. Proceedings of the National Academy of Sciences of the United States of America 98: 10869-108674.

Sorlie, T., Tibshirani, R., Parker, J., Hastie, T., Marron, J. S., Nobel, A., Deng, S., Johnsen, H., Pesich, R., Geisler, S., Demeter, J., Perou, C. M., Lonning, P. E., Brown, P. O., Borresen-Dale, A.-L. and Botstein, D. (2003). Repeated observation of breast tumor subtypes in independent gene expression data sets. Proceedings of the National Academy of Sciences of the United States of America 100: 8418-8423.

~~Soussi, T., Dehouche, K. and Beroud, C. (2000). p53 website and analysis of p53 gene mutations in human cancer: forging a link between epidemiology and carcinogenesis. Human Mutation 15: 105-113.~~

~~Soussi, T., Kato, S., Levy, P. P. and Ishioka, C. (2005). Reassessment of the TP53 Mutation Database in Human Disease by Data Mining With a Library of TP53 Missense Mutations. Human Mutation 25: 6-17.~~

Spentzos, D., Levine, D. A., Ramoni, M. F., Joseph, M., Gu, X., Boyd, J., Libermann, T. A. and Cannistra, S. A. (2004). Gene Expression Signature With Independent Prognostic Significance in Epithelial Ovarian Cancer. Journal of Clinical Oncology 22: 4700-4710.

~~Sporn, M. B. (1996). The war on cancer. Lancet 347: 1377-1381.~~

~~Srivastava, S., Zou, Z. Q., Pirollo, K., Blattner, W. and Chang, E. H. (1990). Germ-line transmission of a mutated p53 gene in a cancer-prone family with Li-Fraumeni syndrome. Nature 348: 747-749.~~

Stein, S., Thomas, E. K., Herzog, B., Westfall, M. D., Rocheleau, J. V., Jackson, R. S., 2nd, Wang, M. and Liang, P. (2004). NDRG1 is necessary for p53-dependent apoptosis. Journal of Biological Chemistry 279: 48930-48940.

Stojadinovic, A., Leung, D. H., Allen, P., Lewis, J. J., Jaques, D. P., Brennan, M. F., Stojadinovic, A., Leung, D. H. Y., Allen, P., Lewis, J. J., Jaques, D. P. and Brennan, M. F. (2002a). Primary adult soft tissue sarcoma: time-dependent influence of prognostic variables. Journal of Clinical Oncology 20: 4344-52.

~~Stojadinovic, A., Leung, D. H., Hoos, A., Jaques, D. P., Lewis, J. J., Brennan, M. F., Stojadinovic, A., Leung, D. H. Y., Hoos, A., Jaques, D. P., Lewis, J. J. and Brennan, M.~~

~~462~~

~~Bibliography~~

~~F. (2002b). Analysis of the prognostic significance of microscopic margins in 2,084 localized primary adult soft tissue sarcomas. Annals of Surgery 235: 424-34.~~

~~Storm, H. H. (1994). Cancers of the soft tissues. Cancer Surveys 20: 197-217.~~

~~Strander, H., Turesson, I. and Cavallin-Stahl, E. (2003). A systematic overview of radiation therapy effects in soft tissue sarcomas. Acta Oncologica. 42: 516-531.~~

Su, A. I., Welsh, J. B., Sapinoso, L. M., Kern, S. G., Dimitrov, P., Lapp, H., Schultz, P. G., Powell, S. M., Moskaluk, C. A., Frierson, H. F., Jr. and Hampton, G. M. (2001). Molecular Classification of Human Carcinomas by Use of Gene Expression Signatures. Cancer Research 61: 7388-7393.

Subramanian, S., West, R. B., Corless, C. L., Ou, W., Rubin, B. P., Chu, K. M., Leung, S. Y., Yuen, S. T., Zhu, S., Hernandez-Boussard, T., Montgomery, K., Nielsen, T. O., Patel, R. M., Goldblum, J. R., Heinrich, M. C., Fletcher, J. A. and van de Rijn, M. (2004). Gastrointestinal stromal tumors (GISTs) with KIT and PDGFRA mutations have distinct gene expression profiles. Oncogene 23: 7780-7790.

~~Suh, C. H., Ordonez, N. G. and Mackay, B. (2000). Malignant fibrous histiocytoma: an ultrastructural perspective. Ultrastructural Pathology 24: 243-250.~~

Sun, Z., Arendt, C. W., Ellmeier, W., Schaeffer, E. M., Sunshine, M. J., Gandhi, L., Annes, J., Petrzilka, D., Kupfer, A., Schwartzberg, P. L. and Littman, D. R. (2000). PKC-theta is required for TCR-induced NF-kappaB activation in mature but not immature T lymphocytes. Nature. 404: 402-407.

~~Suzuki, T., Higgins, P. J. and Crawford, D. R. (2000). Control selection for RNA quantitation. Biotechniques 29: 332-337.~~

Svarvar, C., Bohling, T., Berlin, O., Gustafson, P., Folleras, G., Bjerkehagen, B., Domanski, H. A., Sundby Hall, K., Tukiainen, E., Blomqvist, C. and the Scandinavian Sarcoma Group Leiomyosarcoma Working, G. (2007). Clinical course of nonvisceral soft tissue leiomyosarcoma in 225 patients from the Scandinavian Sarcoma Group. Cancer 109: 282-291.

Taguchi, T., Sonobe, H., Toyonaga, S., Yamasaki, I., Shuin, T., Takano, A., Araki, K., Akimaru, K. and Yuri, K. (2002). Conventional and molecular cytogenetic characterization of a new human cell line, GIST-T1, established from gastrointestinal stromal tumor. Laboratory Investigation 82: 663-665.

Tai, A. L. S., Sham, J. S. T., Xie, D., Fang, Y., Wu, Y.-L., Hu, L., Deng, W., Tsao, G. S. W., Qiao, G.-B., Cheung, A. L. M. and Guan, X.-Y. (2006). Co-overexpression of fibroblast growth factor 3 and epidermal growth factor receptor is correlated with the development of nonsmall cell lung carcinoma. Cancer 106: 146-155.

Taniguchi, M., Nishida, T., Hirota, S., Isozaki, K., Ito, T., Nomura, T., Matsuda, H. and Kitamura, Y. (1999). Effect of c-kit Mutation on Prognosis of Gastrointestinal Stromal Tumors. Cancer Research 59: 4297-4300.

~~463~~

~~Bibliography~~

Tarbe, N., Evtimova, V., Burtscher, H., Jarsch, M., Alves, F. and Weidle, U. H. (2001). Transcriptional profiling of cell lines derived from an orthotopic pancreatic tumor model reveals metastasis-associated genes. Anticancer Research 21: 3221-3228.

Taubert, H., Kappler, M., Bache, M., Bartel, F., Kohler, T., Lautenschlager, C., Blumke, K., Wurl, P., Schmidt, H., Meye, A. and Hauptmann, S. (2005). Elevated expression of survivin-splice variants predicts a poor outcome for soft-tissue sarcomas patients. Oncogene 24: 5258-5261.

Tauchi, T., Nakajima, A., Sashida, G., Shimamoto, T., Ohyashiki, J. H., Abe, K., Yamamoto, K. and Ohyashiki, K. (2002). Inhibition of Human Telomerase Enhances the Effect of the Tyrosine Kinase Inhibitor, Imatinib, in BCR-ABL-positive Leukemia Cells. Clinical Cancer Research 8: 3341-3347.

Taylor, D. S., Cheng, X., Pawlowski, J. E., Wallace, A. R., Ferrer, P. and Molloy, C. J. (1999). Epiregulin is a potent vascular smooth muscle cell-derived mitogen induced by angiotensin II, endothelin-1, and thrombin. Proceedings of the National Academy of Sciences of the United States of America 96: 1633-1638.

~~Teicher, B. A. (1999). Sarcomas. Human Cell Culture. Cancer Cell Lines Part 1. J. W. Masters and B. O. Palsson. Dordrecht, Kluwer Academic Publishers. 1: 1-19.~~

~~Teicher, B. A. (2006). Protein kinase C as a therapeutic target. Clinical Cancer Research 12: 5336-5345.~~

Temple, W. J., Temple, C. L., Arthur, K., Schachar, N. S., Paterson, A. H. and Crabtree, T. S. (1997). Prospective cohort study of neoadjuvant treatment in conservative surgery of soft tissue sarcomas. Annals of Surgical Oncology. 4: 586-590. ten Heuvel, S. E., Hoekstra, H. J., van Ginkel, R. J., Bastiaannet, E. and Suurmeijer, A. J. (2007). Clinicopathologic prognostic factors in myxoid liposarcoma: a retrospective study of 49 patients with long-term follow-up. Annals of Surgical Oncology 14: 222- 229.

Terry, J., Saito, T., Subramanian, S., Ruttan, C., Antonescu, C. R., Goldblum, J. R., Downs-Kelly, E., Corless, C. L., Rubin, B. P., van de Rijn, M., Ladanyi, M. and Nielsen, T. O. (2007). TLE1 as a diagnostic immunohistochemical marker for synovial sarcoma emerging from gene expression profiling studies. American Journal of Surgical Pathology 31: 240-246.

Thelen, P., Burfeind, P., Grzmil, M., Voigt, S., Ringert, R. H. and Hemmerlein, B. (2004). cDNA microarray analysis with amplified RNA after isolation of intact cellular RNA from neoplastic and non-neoplastic prostate tissue separated by laser microdissections. International Journal of Oncology 24: 1085-1092.

Thellin, O., Zorzi, W., Lakaye, B., De Borman, B., Coumans, B., Hennen, G., Grisar, T., Igout, A. and Heinen, E. (1999). Housekeeping genes as internal standards: use and limits. Journal of Biotechnology 75: 291-5.

~~464~~

~~Bibliography~~

~~Thiery, J. P. (2002). Epithelial-mesenchymal transitions in tumour progression. Nature Reviews Cancer 2: 442-454.~~

Thomas, C. Y., Chouinard, M., Cox, M., Parsons, S., Stallings-Mann, M., Garcia, R., Jove, R. and Wharen, R. (2003). Spontaneous activation and signaling by overexpressed epidermal growth factor receptors in glioblastoma cells. International Journal of Cancer 104: 19-27.

Thomas, D. G., Giordano, T. J., Sanders, D., Biermann, J. S. and Baker, L. (2002). Absence of HER2/neu Gene Expression in Osteosarcoma and Skeletal Ewing's Sarcoma. Clinical Cancer Research 8: 788-793.

Thomas, D. G., Giordano, T. J., Sanders, D., Biermann, S., Sondak, V. K., Trent, J. C., Yu, D., Pollock, R. E. and Baker, L. (2005). Expression of receptor tyrosine kinases epidermal growth factor receptor and HER-2/neu in synovial sarcoma. Cancer 103: 830-838.

~~Thorpe, P. E. (2004). Vascular Targeting Agents as Cancer Therapeutics. Clinical Cancer Research 10: 415-427.~~

~~Timar, J. and Paterson, H. (1990). Localization and production of proteoglycans by HT1080 cell lines with altered N-ras expression. Cancer Letters. 53: 145-50.~~

Timmerman, L. A., Grego-Bessa, J., Raya, A., Bertran, E., Perez-Pomares, J. M., Diez, J., Aranda, S., Palomo, S., McCormick, F., Izpisua-Belmonte, J. C. and de la Pompa, J. L. (2004). Notch promotes epithelial-mesenchymal transition during cardiac development and oncogenic transformation. Genes & Development 18: 99-115.

~~Tomescu, O. and Barr, F. G. (2001). Chromosomal translocations in sarcomas: prospects for therapy. Trends in Molecular Medicine 7: 554-559.~~

Torhorst, J., Bucher, C., Kononen, J., Haas, P., Zuber, M., Kochli, O. R., Mross, F., Dieterich, H., Moch, H., Mihatsch, M. J., Kallioniemi, O. P. and Sauter, G. (2001). Tissue Microarrays for Rapid Linking of Molecular Changes to Clinical Endpoints. American Journal of Pathology 159: 2249-2256.

~~Tornillo, L. and Terracciano, L. M. (2006). An update on molecular genetics of gastrointestinal stromal tumours. Journal of Clinical Pathology 59: 557-563.~~

Torring, N., Jorgensen, P. E., Sorensen, B. S. and Nexo, E. (2000). Increased expression of heparin binding EGF (HB-EGF), amphiregulin, TGF alpha and epiregulin in androgen-independent prostate cancer cell lines. Anticancer Research 20: 91-95.

Toyoda, H., Komurasaki, T., Uchida, D. and Morimoto, S. (1997). Distribution of mRNA for human epiregulin, a differentially expressed member of the epidermal growth factor family. Biochemical Journal 326: 69-75.

~~465~~

~~Bibliography~~

Trassard, M., Le Doussal, V., Hacene, K., Terrier, P., Ranchere, D., Guillou, L., Fiche, M., Collin, F., Vilain, M. O., Bertrand, G., Jacquemier, J., Sastre-Garau, X., Bui, N. B., Bonichon, F. and Coindre, J. M. (2001). Prognostic factors in localized primary synovial sarcoma: a multicenter study of 128 adult patients. Journal of Clinical Oncology 19: 525-534.

Trautinger, F., Kokesch, C., Klosner, G., Knobler, R. M. and Kindas-Mugge, I. (1999). Expression of the 72-kD heat shock protein is induced by ultraviolet A radiation in a human fibrosarcoma cell line. Experimental Dermatology. 8: 187-92.

Trojani, M., Contesso, G., Coindre, J. M., Rouesse, J., Bui, N. B., de Mascarel, A., Goussot, J. F., David, M., Bonichon, F. and Lagarde, C. (1984). Soft-tissue sarcomas of adults; study of pathological prognostic variables and definition of a histopathological grading system. International Journal of Cancer 33: 37-42.

Trovik, C. S., Bauer, H. C., Alvegard, T. A., Anderson, H., Blomqvist, C., Berlin, O., Gustafson, P., Saeter, G. and Walloe, A. (2000). Surgical margins, local recurrence and metastasis in soft tissue sarcomas: 559 surgically-treated patients from the Scandinavian Sarcoma Group Register. European Journal of Cancer 36: 710-716.

Tsao, M.-S., Sakurada, A., Cutz, J.-C., Zhu, C.-Q., Kamel-Reid, S., Squire, J., Lorimer, I., Zhang, T., Liu, N., Daneshmand, M., Marrano, P., da Cunha Santos, G., Lagarde, A., Richardson, F., Seymour, L., Whitehead, M., Ding, K., Pater, J. and Shepherd, F. A. (2005a). Erlotinib in Lung Cancer -- Molecular and Clinical Predictors of Outcome. New England Journal of Medicine 353: 133-144.

Tsao, M. S., Sakurada, A., Cutz, J. C., Zhu, C. Q., Kamel-Reid, S., Squire, J., Lorimer, I., Zhang, T., Liu, N., Daneshmand, M., Marrano, P., da Cunha Santos, G., Lagarde, A., Richardson, F., Seymour, L., Whitehead, M., Ding, K., Pater, J. and Shepherd, F. A. (2005b). Erlotinib in lung cancer - molecular and clinical predictors of outcome. New England Journal of Medicine 353: 133-144.

~~Tsujimoto, M., Aozasa, K., Ueda, T., Morimura, Y., Komatsubara, Y. and Doi, T. (1988). Multivariate analysis for histologic prognostic factors in soft tissue sarcomas. Cancer 62: 994-998.~~

Tsujimoto, Y., Yunis, J., Onorato-Showe, L., Erikson, J., Nowell, P. C. and Croce, C. M. (1984). Molecular cloning of the chromosomal breakpoint of B-cell lymphomas and leukemias with the t(11;14) chromosome translocation. Science 224: 1403-1406.

Tsutsui, T., Fujino, T., Kodama, S., Tainsky, M. A., Boyd, J. and Barrett, J. C. (1995). Aflatoxin B1-induced immortalization of cultured skin fibroblasts from a patient with Li-Fraumeni syndrome. Carcinogenesis 16: 25-34.

Tsutsui, T., Kumakura, S., Tamura, Y., Tsutsui, T. W., Sekiguchi, M., Higuchi, T. and Barrett, J. C. (2003). Immortal, telomerase-negative cell lines derived from a Li- Fraumeni syndrome patient exhibit telomere length variability and chromosomal and minisatellite instabilities. Carcinogenesis 24: 953-965.

~~466~~

~~Bibliography~~

Tsutsui, T., Kumakura, S., Yamamoto, A., Kanai, H., Tamura, Y., Kato, T., Anpo, M., Tahara, H. and Barrett, J. C. (2002). Association of p16(INK4a) and pRb inactivation with immortalization of human cells. Carcinogenesis 23: 2111-7.

Tsutsui, T., Tanaka, Y., Matsudo, Y., Hasegawa, K., Fujino, T., Kodama, S. and Barrett, J. C. (1997). Extended lifespan and immortalization of human fibroblasts induced by X- ray irradiation. Molecular Carcinogenesis 18: 7-18.

Tuveson, D. A., Willis, N. A., Jacks, T., Griffin, J. D., Singer, S., Fletcher, C. D., Fletcher, J. A. and Demetri, G. D. (2001). STI571 inactivation of the gastrointestinal stromal tumor c-KIT oncoprotein: biological and clinical implications. Oncogene 20: 5054-5058.

Ulloa, L. and Messmer, D. (2006). High-mobility group box 1 (HMGB1) protein: friend and foe. Cytokine & Growth Factor Reviews 17: 189-201. van 't Veer, L. J., Dai, H., van de Vijver, M. J., He, Y. D., Hart, A. A., Mao, M., Peterse, H. L., van der Kooy, K., Marton, M. J., Witteveen, A. T., Schreiber, G. J., Kerkhoven, R. M., Roberts, C., Linsley, P. S., Bernards, R. and Friend, S. H. (2002). Gene expression profiling predicts clinical outcome of breast cancer. Nature 415: 530-536. van Belzen, N., Dinjens, W. N., Diesveld, M. P., Groen, N. A., van der Made, A. C., Nozawa, Y., Vlietstra, R., Trapman, J. and Bosman, F. T. (1997). A novel gene which is up-regulated during colon epithelial cell differentiation and down-regulated in colorectal neoplasms. Laboratory Investigation 77: 85-92. van de Rijn, M. and Rubin, B. P. (2002). Gene expression studies on soft tissue tumors. American Journal of Pathology. 161: 1531-1534. van der Flier, A. and Sonnenberg, A. (2001). Function and interactions of integrins. Cell & Tissue Research 305: 285-298. van Oosterom, A. T., Judson, I., Verweij, J., Stroobants, S., Donato di Paola, E., Dimitrijevic, S., Martens, M., Webb, A., Sciot, R., Van Glabbeke, M., Silberman, S., Nielsen, O. S., European Organisation for, R., Treatment of Cancer Soft, T. and Bone Sarcoma, G. (2001). Safety and efficacy of imatinib (STI571) in metastatic gastrointestinal stromal tumours: a phase I study. Lancet 358: 1421-1423. van Oosterom, A. T., Judson, I. R., Verweij, J., Stroobants, S., Dumez, H., Donato di Paola, E., Sciot, R., Van Glabbeke, M., Dimitrijevic, S., Nielsen, O. S., European Organisation for, R., Treatment of Cancer Soft, T. and Bone Sarcoma, G. (2002). Update of phase I study of imatinib (STI571) in advanced soft tissue sarcomas and gastrointestinal stromal tumors: a report of the EORTC Soft Tissue and Bone Sarcoma Group. European Journal of Cancer 38.

~~Vandesompele, J., De Paepe, A. and Speleman, F. (2001). Elimination of Primer±Dimer Artifacts and Genomic Coamplification Using a Two-Step SYBR Green I Real-Time RT-PCR. Analytical Biochemistry.~~

~~467~~

~~Bibliography~~

Vandesompele, J., De Preter, K., Pattyn, F., Poppe, B., Van Roy, N., De Paepe, A. and Speleman, F. (2002). Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes. Genome Biology 3: research0034.1–0034.11.

Varley, J. M., Attwooll, C., White, G., McGown, G., Thorncroft, M., Kelsey, A. M., Greaves, M., Boyle, J. and Birch, J. M. (2001). Characterization of germline TP53 splicing mutations and their genetic and functional analysis. Oncogene 20: 2647-2654.

~~Varley, J. M., Evans, D. G. and Birch, J. M. (1997). Li-Fraumeni syndrome--a molecular and clinical review. British Journal of Cancer 76: 1-14.~~

Verweij, J., van Oosterom, A., Blay, J. Y., Judson, I., Rodenhuis, S., van der Graaf, W., Radford, J., Le Cesne, A., Hogendoorn, P. C., di Paola, E. D., Brown, M. and Nielsen, O. S. (2003). Imatinib mesylate (STI-571 Glivec, Gleevec) is an active agent for gastrointestinal stromal tumours, but does not yield responses in other soft-tissue sarcomas that are unselected for a molecular target. Results from an EORTC Soft Tissue and Bone Sarcoma Group phase II study. European Journal of Cancer 39: 2006-2011.

~~Vivanco, I. and Sawyers, C. L. (2002). The phosphatidylinositol 3-Kinase AKT pathway in human cancer. Nature Reviews. Cancer. 2: 489-501.~~

Vogelstein, B., Fearon, E. R., Hamilton, S. R., Kern, S. E., Preisinger, A. C., Leppert, M., Nakamura, Y., White, R., Smits, A. M. and Bos, J. L. (1988). Genetic alterations during colorectal-tumor development. New England Journal of Medicine 319: 525-532. von Mehren, M. (2007). New therapeutics for soft-tissue sarcomas in adults. Oncology 21: 123-126. von Wasielewski, R., Mengel, M., Wiese, B., Rudiger, T., Muller-Hermelink, H. K. and Kreipe, H. (2002). Tissue array technology for testing interlaboratory and interobserver reproducibility of immunohistochemical estrogen receptor analysis in a large multicenter trial. American Journal of Clinical Pathology. 118: 675-682.

Vonderheide, R. H., Domchek, S. M., Schultze, J. L., George, D. J., Hoar, K. M., Chen, D. Y., Stephans, K. F., Masutomi, K., Loda, M., Xia, Z., Anderson, K. S., Hahn, W. C. and Nadler, L. M. (2004). Vaccination of cancer patients against telomerase induces functional antitumor CD8+ T lymphocytes. Clinical Cancer Research 10: 828-839.

Wabitsch, M., Bruderlein, S., Melzner, I., Braun, M., Mechtersheimer, G. and Moller, P. (2000). LiSa-2, a novel human liposarcoma cell line with a high capacity for terminal adipose differentiation. International Journal of Cancer 88: 889-894.

Wai, D. H., Schaefer, K. L., Schramm, A., Korsching, E., Van Valen, F., Ozaki, T., Boecker, W., Schweigerer, L., Dockhorn-Dworniczak, B. and Poremba, C. (2002). Expression analysis of pediatric solid tumor cell lines using oligonucleotide microarrays. International Journal of Oncology 20: 441-451.

~~468~~

~~Bibliography~~

Wall, S. J. and Edwards, D. R. (2002). Quantitative Reverse Transcription±Polymerase Chain Reaction (RT-PCR): A Comparison of Primer-Dropping, Competitive, and Real- Time RT-PCR. Analytical Biochemistry 300: 269-273.

~~Wanebo, H. J., Temple, W. J., Popp, M. B., Constable, W., Aron, B. and Cunningham, S. L. (1995). Preoperative regional therapy for extremity sarcoma. A tricenter update. Cancer 75: 2299-2306.~~

Wang, L., Habuchi, T., Takahashi, T., Mitsumori, K., Kamoto, T., Kakehi, Y., Kakinuma, H., Sato, K., Nakamura, A., Ogawa, O. and Kato, T. (2002). Cyclin D1 gene polymorphism is associated with an increased risk of urinary bladder cancer. Carcinogenesis 23: 257-264.

Wang, R., Lu, Y. J., Fisher, C., Bridge, J. A. and Shipley, J. (2001). Characterization of chromosome aberrations associated with soft-tissue leiomyosarcomas by twenty-four- color karyotyping and comparative genomic hybridization analysis. Genes, Chromosomes & Cancer 31: 54-64.

Wang, R., Titley, J., Lu, Y., Summersgill, B., Bridge, J., Fisher, C. and Shipley, J. (2003). Loss of 13q14-q21 and gain of 5p14-pter in the progression of leiomyosarcoma. Modern Pathology. 16: 778-785.

Wang, Z., Wang, F., Wang, W. Q., Gao, Q., Wei, W. L., Yang, Y. and Wang, G. Y. (2004). Correlation of N-myc downstream-regulated gene 1 overexpression with progressive growth of colorectal neoplasm. World Journal of Gastroenterology 10: 550- 554.

Wardelmann, E., Neidt, I., Bierhoff, E., Speidel, N., Manegold, C., Fischer, H. P., Pfeifer, U. and Pietsch, T. (2002). c-kit mutations in gastrointestinal stromal tumors occur preferentially in the spindle rather than in the epithelioid cell variant. Modern Pathology 15: 125-136.

Wei-wei, L., Qiu-liang, W., Guo-hao, W., Zhi-hua, C. and Zong-yuan, Z. (2005). Clinicopathologic features, treatment, and prognosis of postirradiation osteosarcoma in patients with nasopharyngeal cancer. Laryngoscope 115: 1574-1579.

~~Wen, Z., Zhong, Z. and Darnell, J. E., Jr. (1995). Maximal activation of transcription by Stat1 and Stat3 requires both tyrosine and serine phosphorylation. Cell 82: 241-250.~~

Weng, W. H., Wejde, J., Ahlen, J., Pang, S. T., Lui, W. O. and Larsson, C. (2004). Characterization of large chromosome markers in a malignant fibrous histiocytoma by spectral karyotyping, comparative genomic hybridization (CGH), and array CGH. Cancer Genetics & Cytogenetics 150: 27-32.

West, R. B., Corless, C. L., Chen, X., Rubin, B. P., Subramanian, S., Montgomery, K., Zhu, S., Ball, C. A., Nielsen, T. O., Patel, R., Goldblum, J. R., Brown, P. O., Heinrich, M. C. and van de Rijn, M. (2004). The novel marker, DOG1, is expressed ubiquitously in gastrointestinal stromal tumors irrespective of KIT or PDGFRA mutation status. American Journal of Pathology 165: 107-113.

~~469~~

~~Bibliography~~

~~White, L. K., Wright, W. E. and Shay, J. W. (2001). Telomerase inhibitors. Trends in Biotechnology 19: 114-120.~~

Wilda, M., Ghaffari-Tabrizi, N., Reisert, I., Utermann, G., Baier, G. and Hameister, H. (2001). Protein kinase C isoenzyme: selective expression pattern of protein kinase C- theta during mouse development. Mechanisms of Development 103: 197-200.

Wilson, A. N., Davis, A., Bell, R. S., O'Sullivan, B., Catton, C., Madadi, F., Kandel, R. and Fornasier, V. L. (1994). Local control of soft tissue sarcoma of the extremity: the experience of a multidisciplinary sarcoma group with definitive surgery and radiotherapy. European Journal of Cancer. 6: 746-751.

Wilson, R. B., Crowe, P. J., Fisher, R., Hook, C. and Donnellan, M. J. (1999). Extremity soft tissue sarcoma: factors predictive of local recurrence and survival. Australian and New Zealand Journal of Surgery 69: 344-349.

Wright, W. E., Shay, J. W. and Piatyszek, M. A. (1995). Modifications of a telomeric repeat amplification protocol (TRAP) result in increased reliability, linearity and sensitivity. Nucleic Acids Research 23: 3794-3795.

Wurl, P., Kappler, M., Meye, A., Bartel, F., Kohler, T., Lautenschlager, C., Bache, M., Schmidt, H. and Taubert, H. (2002). Co-expression of survivin and TERT and risk of tumour-related death in patients with soft-tissue sarcoma. The Lancet 359: 943-945.

Wurl, P., Meye, A., Lautenschlager, C., Schmidt, H., Bache, M., Kalthoff, H., Schonfelder, M., Rath, F. W. and Taubert, H. (1999). Clinical relevance of pRb and p53 co-overexpression in soft tissue sarcomas. Cancer Letters 139: 159-165.

Xiaojuan Sun, L. W., Johan Lidén, Gao Hui, Karin Dahlman-Wright, Anders Hjerpe, Katalin Dobra, (2005). Molecular characterization of tumour heterogeneity and malignant mesothelioma cell differentiation by gene profiling. The Journal of Pathology 207: 91-101.

Yang, J. C., Chang, A. E., Baker, A. R., Sindelar, W. F., Danforth, D. N., Topalian, S. L., DeLaney, T., Glatstein, E., Steinberg, S. M., Merino, M. J. and Rosenberg, S. A. (1998). Randomized prospective study of the benefit of adjuvant radiation therapy in the treatment of soft tissue sarcomas of the extremity. Journal of Clinical Oncology 16: 197-203.

Yang, J. L., Qu, X. J., Hayes, V. M., Brenner, P. C., Russell, P. J. and Goldstein, D. (2007). Erlotinib (OSI-774)-induced inhibition of transitional cell carcinoma of bladder cell line growth is enhanced by interferon-α. BJU International 99: 1539-1545.

Yang, Y. H., Dudoit, S., Luu, P., Lin, D. M., Peng, V., Ngai, J. and Speed, T. P. (2002). Normalization for cDNA microarray data: a robust composite method addressing single and multiple slide systematic variation. Nucleic Acids Research. 30: e15.

~~470~~

~~Bibliography~~

~~Yang, Y. H. and Speed, T. (2002). Design issues for cDNA microarray experiments. Nature Reviews Genetics. 3: 579-588.~~

~~Yarden, Y. (2001). The EGFR family and its ligands in human cancer. signalling mechanisms and therapeutic opportunities. European Journal of Cancer 37: S3-8.~~

~~Yasko, A. W. (2002). Biopsy Techniques. Atlas of Clinical Oncology: Soft Tissue Sarcomas. R. E. Pollock. Hamilton, Ontario, BC Decker Inc: 70-79.~~

Yeager, T. R., Neumann, A. A., Englezou, A., Huschtscha, L. I., Noble, J. R. and Reddel, R. R. (1999). Telomerase-negative immortalized human cells contain a novel type of promyelocytic leukemia (PML) body. Cancer Research 59: 4175-4179.

Yeoh, E. J., Ross, M. E., Shurtleff, S. A., Williams, W. K., Patel, D., Mahfouz, R., Behm, F. G., Raimondi, S. C., Relling, M. V., Patel, A., Cheng, C., Campana, D., Wilkins, D., Zhou, X., Li, J., Liu, H., Pui, C. H., Evans, W. E., Naeve, C., Wong, L. and Downing, J. R. (2002). Classification, subtype discovery, and prediction of outcome in pediatric acute lymphoblastic leukemia by gene expression profiling. Cancer Cell 1: 133-143.

Yoo, J., Park, S. Y., Kang, S. J., Shim, S. I. and Kim, B. K. (2002). Altered expression of G1 regulatory proteins in human soft tissue sarcomas. Archives of Pathology & Laboratory Medicine 126: 567-573.

Yoon, S. S., Segal, N. H., Park, P. J., Detwiller, K. Y., Fernando, N. T., Ryeom, S. W., Brennan, M. F. and Singer, S. (2006). Angiogenic profile of soft tissue sarcomas based on analysis of circulating factors and microarray gene expression. Journal of Surgical Research 135: 282-290.

~~Yu, L., Hebert, M. C. and Zhang, Y. E. (2002). TGF-beta receptor-activated p38 MAP kinase mediates Smad-independent TGF-beta responses. EMBO Journal 21: 3749-3759.~~

~~Yu, X., Miyamoto, S. and Mekada, E. (2000). Integrin alpha 2 beta 1-dependent EGF receptor activation at cell-cell contact sites. Journal of Cell Science 113: 2139-2147.~~

Yudoh, K., Kanamori, M., Ohmori, K., Yasuda, T., Aoki, M. and Kimura, T. (2001). Concentration of vascular endothelial growth factor in the tumour tissue as a prognostic factor of soft tissue sarcomas. British Journal of Cancer. 84: 1610-1615.

~~Yuen, T., Wurmbach, E., Pfeffer, R. L., Ebersole, B. J. and Sealfon, S. C. (2002). Accuracy and calibration of commercial oligonucleotide and custom cDNA microarrays. Nucleic Acids Research 30: 15.~~

Zagars, G. K., Ballo, M. T., Pisters, P. W., Pollock, R. E., Patel, S. R. and Benjamin, R. S. (2003a). Prognostic factors for disease-specific survival after first relapse of soft- tissue sarcoma: analysis of 402 patients with disease relapse after initial conservative surgery and radiotherapy. International Journal of Radiation Oncology, Biology, Physics. 57: 739-747.

~~471~~

~~Bibliography~~

Zagars, G. K., Ballo, M. T., Pisters, P. W., Pollock, R. E., Patel, S. R., Benjamin, R. S. and Evans, H. L. (2003b). Prognostic factors for patients with localized soft-tissue sarcoma treated with conservation surgery and radiation therapy: an analysis of 1225 patients. Cancer. 97: 2530-2543.

Zandi, R., Larsen, A. B., Andersen, P., Stockhausen, M. T., Poulsen, H. S., Zandi, R., Larsen, A. B., Andersen, P., Stockhausen, M.-T. and Poulsen, H. S. (2007). Mechanisms for oncogenic activation of the epidermal growth factor receptor. Cellular Signalling 19: 2013-23.

~~Zavadil, J., Cermak, L., Soto-Nieves, N. and Bottinger, E. P. (2004). Integration of TGF-beta/Smad and Jagged1/Notch signalling in epithelial-to-mesenchymal transition. EMBO Journal 23: 1155-1165.~~

~~Zhang, H. S., Postigo, A. A. and Dean, D. C. (1999). Active transcriptional repression by the Rb-E2F complex mediates G1 arrest triggered by p16INK4a, TGFbeta, and contact inhibition. Cell 97: 53-61.~~

~~Zhong, S., Salomoni, P. and Pandolfi, P. P. (2000). The transcriptional role of PML and the nuclear body. Nature Cell Biology 2: E85-E90.~~

~~472~~