Understanding Childhood Cancer: Morphology, Molecular Genetics, and Models

1
Understanding Childhood CancerMorphology,
Molecular Genetics, and Models
The Third M. Daria Haust Lecture Queens
University April 26, 2005

• Timothy J. Triche, MD, PhD
• Childrens Hospital Los Angeles
• Keck School of Medicine at USC

2
Learning Objectives

• Discuss the critical role of morphology in
  pathologic diagnosis, as well as its limitations
• Document the role of gene-based methods for
  childhood cancer diagnosis prognosis
• Illustrate the implications of molecular methods
  for improved clinical relevance of pathologic
  diagnosis in general
• Predict how knowledge of pathogenic genes could
  lead to targeted therapy using mouse models of
  childhood cancer

3
Overview

• Morphology the current gold standard for
  pathologic diagnosis-
• HE or cytology
• EM
• IHC
• Diagnosis drives therapy
• Is it Cancer?
• What is the preferred therapy- surgery,
  chemo-therapy, radiation, new agents?
• Current and especially future treatment will
  depend heavily on analysis of genetic factors
• Mutated translocated genes
• Patterns of gene expression
• Structural genomic alterations
• Druggable gene targets

4
Molecular Diagnostic Market
Source Technology Review April 2005
5
The Current Diagnostic Standard
Rhabdomyosarcoma
Osteosarcoma
Histology
Ewings sarcoma
Fibrosarcoma
6
But When Morphology Fails?

• Histogenesis?
• Phenotype?
• Genomic alterations?
• Malignancy?
• Therapeutic response?
• Metastases?
• Outcome?

7
Single-Gene Cancer Diagnostics

• MYCN amplification in neuroblastoma
• Clonal Ig gene rearrangements in leukemia
  lymphoma
• P53 mutations in Li-Fraumeni syndrome
• RB gene mutations in familial retinoblastoma
• Gene translocations in leukemia
• Gene translocations in sarcomas

8
Cytogenetic Diagnostics
9
Ewings Sarcoma Translocation
chr 11 chr 22
EWS
EWS FLI-1
FLI-1
Normal Der Normal Der
10
Childhood Cancer is At Heart a Genetic Disease

• PAX/FKHR fusion gene found only in Alveolar RMS
• EWS/ets defines Ewings/pPNETs, including EOE in
  IRS
• SYT/SSX defines synoviosarcoma
• TEL/TRK defines congenital fibrosarcoma

Sarcoma Fusion Gene Alveolar
RMS PAX3,7/FKHR Embryonal RMS None
(LOI,LOH) Ewings sarcoma EWS/ets
Synoviosarcoma SYT/SSX1,2 Fibrosarcoma
(cong) ETV/TRKC Desmoplastic RCT EWS/WT-1 Soft
Part Melanoma EWS/ATF Liposarcoma TLS/CHOP
11
Ewings EWS/Fli-1 Transcripts
1
656
EWS
RNA BD
COO
H
NH
2
Type 1 Fusion
EWS /
COO
H
ETS D
NH
2
Fli-1
Human
1
452
COO
H
ETS D
ETS D
NH
2
Fli-1
12
Chimeric Gene Transcript PCR

• Virtually all Ewings Tumors express a variant of
  EWS-FLI-1 or an EWS-ets chimeric gene

Type 3 Type 2 Type 1
13
Ewings Interphase FISH
14
Single Gene Diagnostics Limitations

• Useful for diagnosis, but not 100 accurate
• Less reliable for estimating prognosis
• MYCN in neuroblastoma is a good example

15
Gene Expression Profiling Basic Process
RNA
AAAAA
Tissue selection
cDNA
AAAAA
T7-TTTTT
labeled cRNA
Fragment cRNA mix




Bioinformatics
Information









Data
Hybridize wash
Analyze
Scan
16
The Goal

• Assess the entire expressed genome in cancer
• Identify candidate genes that denote class
  (diagnosis), prognosis, and potential therapeutic
  targets
• Develop targeted molecular therapies

17
Bioinformatics Challenges
18
(No Transcript)
19
(No Transcript)
20
Molecular ClassificationbyGene Expression
Profilingis a PowerfulDiagnostic Tool
21
SRCT Dx by Hierarchical Clustering
22
Gene Expression Profiling of 81 Childhood Cancers
Genetrix 3D-scatter plot using principal
components
23
Independent Validation

• 600 cases of sarcoma
• 22,000 gene profiling
• Multi-dimensional scaling

Ewings Osteo RMS NRSTS
24
Prognostic PredictionbyGene Expression
Profilingis alsoFeasible
25
Osteosarcoma Clinical PrognosisResponse to
Chemotherapy
R
NR
26
Outcome Prediction
Minimum
N39
Alive Dead
Error Rate 1 case in 39 (3.7)
27
Poor Prognosis Genes
The 10 worst

1. TRC8 Patched related protein, WNT TGF?
   inhibitor
2. ABC ATP-binding cassette, MRP-related (drug
   resistance)
3. Tomosyn-like
4. HDAC5 Histone deacetylase 5, gene transcription
   cont.
5. SMNP Survival motor neuron pseudogene
6. PTPRR protein tyrosine phosphatase, MAPK
   localizer
7. PRPH Peripherin, memb.-spanning intermed.
   filament
8. NFATC4 Nuclear factor of activated T-cells, TF
9. DOCK3 Dedicator of cyto-kinesis 3
10. Timeless homolog (Drosophila)- circadian rhythm
    reg

28
Favorable Prognosis Genes
The 10 best

1. IF I factor (complement-inactivating serine
   protease)
2. GNAI2 G-protein complex inhibitor
3. LAPTM5 Lysosomal-associated membrane protein 5
4. ZFPL1 Zinc finger protein-like 1
5. PRG1 Proteoglycan 1, secretory granule
6. HLA-E Major histocompatibility complex, class I,
   E
7. PTPRC, protein tyrosine phosphatase, receptor
   type C
8. Zx53d03.r1 (no annotation)
9. WNT 5A Wingless-type MMTV integration site
   family
10. HCK Hemopoietic cell kinase, non-receptor TYR
    kinase

29
Prognosis10 Genes Predictive of Survival
Worst
Best
30
Rhabdomyosarcoma Subtype Diagnosis

• 2 major forms Embryonal Alveolar
• Prognosis differs
• Embryonal 85 survival
• Alveolar 50 survival
• Differing genetic defects described
• Embryonal- loss of 1 parental allele on
  chromosome 11
• Alveolar- recurring gene translocation in most
  cases

31
RMS Prognosis by Histologic Type
IRS IV Data
32
Alveolar RMS Translocation
chr 2
chr 13
n d
n d
n d
13q14
2q35
33
Gene Fusions in 171 IRS-IV Cases
Fusion negative
PAX7
PAX3
n
Histology
93
0
0
93
ERMS, Botryoid Spindle, Undiff.
18 (23)
17 (22)
43 (55)
78
Alveolar
34
Prognosis by PAX Fusion Gene
35
Distance Matrix (Metaclustering)
Alveolar
Embryonal
Other
36
Rhabdomyosarcoma Classification

• Unsupervised Clustering X100
• Optimal genes extracted X100
• 39 optimal genes chosen (100/100)
• 3D PCA analysis
• 5 ARMS PAX neg always cluster with ERMS

Alveolar RMS
Other sarcoma
Spindle Botryoid RMS
Embryonal RMS
37
Misclassified ARMS
No slides available on 1 case other 4
illustrated here
38
Typical Embryonal RMS
39
Re-clustering (corrected Dx)
P-F Alveolar RMS
Embryonal RMS
Non-RMS STS
40
QRT PCR of ARMS P-F Expression
ERMS

• Analysis of additional ARMS case
• Translocation
• Translocation , low expressors
• Only high expressors (red) cluster together
• Others (black, blue) resemble ERMS (gray)

ARMS
41
Embryonal vs. Alveolar RMS

• Embryonal always lacks a PAX-FKHR gene
  translocation
• Embryonal RMS, hepatoblastoma, Wilms tumor
  show loss of heterozygosity (LOH) on 11p1.3-pter
• Alveolar shows little (or no) 11p LOH

42
LOH by 10K SNP Chip in RMS
Embryonal RMS Alveolar RMS
43
Genes Responsible for Class ID
16 genes
Consensus minimum
Error 11.2
44
IHC Confirmation of Molecular Dx
TFAP2ß
HMGA2
45
Corrected Gene-Based Classification
Alveolar RMS
Embryonal RMS
Other sarcoma
46
ERMS Classified as Other Sarcoma5 of 6 Appear
Myogenic, but Atypical for ERMS
47
RMS- Molecular Class Distinction
48
Structural Genomic Changes
Not myogenic! Myogenic!
Allelic loss / gain on chr 10, 11 in ERMS, not
ARMS or NRSTS
49
Summary of Genomic LOH in Sarcomas
True ARMS, ERMS, NRSTS are structurally
distinct
ARMS
ERMS
NRSTS
Method 100K SNP Chip analysis of allelic loss
gain
50
Conclusion

• All true alveolar RMS possess a PAX-FKHR
  translocation thus is a genetic, not
  morphologic, entity
• Distinction between myogenic non-myogenic STS
  can be made by GEPs
• Structural genomic differences exist between the
  three major STSs of childhood

51
Therapeutic Target Identification

• Therapy based on specific gene targets with known
  biologic function for which interventions may be
  available

52
FGFR4 Expression in Sarcomas
NB ERMS ARMS-3 ARMS-7 EFTs-tc
EFTs-tumor
53
Receptor Tyrosine Kinases (RTKs)
EGFR HER2 HER3 HER4
INSR IGF-1R IRR
PDGFR CSF-1R KIT FLK2
FLT1 FLK1 FLT4
FGFR1 FGFR2 FGFR3 FGFR4
CCK4
MET RON
TRKA TRKB TRKC
AXL MER SKY EYK
TIE TEK
EPH ECK HEK SEK1 EHK1 EHK2 MDK1 EEK ELK NUK
RYK
RET
LTK ALK
Many others..
54
FGFR4 Expression vs. Survival
55
Caution Single Biomarkers Rarely Suffice as
Single Class Discriminators

• CA125, AFP, PSA, etc. all show both false
  positives and false negatives
• Even immunohistochemical markers are used
  judiciously in groups to reduce false positive
  and negative calls
• Multi-marker panels are generally more reliable

56
RMS Survival by Clinical Group
57
Metagene Prediction IRS Risk Group
Expression level of 48 genes only!

• Composite of gt 6 covariates
• Age
• Stage
• Site
• Histology
• Special sites
• Status post-surgery

best
intermediate
worst
Metagene
58
What About Therapeutic Targets?

• Genes responsible for behavior of these tumors
  are readily identified
• Biologic role suggests potential therapeutic
  targets
• New agents can be developed for specific gene
  targets

59
Prognosis Metagene Opposing Biological Pathways
Functional Annotation of Genes associated with
Good and Poor Prognosis
Poorly Myogenic RMS Poor Prognosis!
Highly Myogenic RMS Good Prognosis

• muscle genes plt0.0028
• MRFs, myosins,
• p38 signaling
• MEKK3, MAPKAPK3

• neurogenesis plt0.001
• angiogenesis plt0.02
• HIF2a, VEGF, IL8
• JNK signaling
• MAP3K5, MAPK8IP
• chromosome 16q plt0.029

60
siRNA Mediated Gene Silencing

• Advantage
• Potent, complete destruction of specific mRNAs

Disadvantage A potential targeted therapeutic
with few side effects,but limited to in vitro use
to date, due to extreme lability with systemic
administration, due to ubiquitous RNAse activity.
Adapted from Novina, 2004
61
Targeted Delivery of siRNA Nanoparticle-based
encapsulation
50 nm
Assembly of the non-targeted and targeted
particles
Components of the delivery system
62
Experimental Design
Xenogen imaging
Luciferase ATP
Luciferin O2 ---gt Oxyluciferin CO2
LIGHT
5x106 Cells per mice
CDP/siRNA
NOD/scid 4-6wk
63
In Vitro Efficacy of CDP/siRNA Particles
Knockdown of the EWS-FLI1 Gene
(adapted from Dohjima et al., Molecular Therapy,
2003)
64
Establishment of a metastatic EFT model
65
(No Transcript)
66
(No Transcript)
67
Growth Curves for Engrafted Tumors
68
Conclusions

• Gene-based diagnostics continue to expand into
  routine pathologic diagnosis
• Whole-genome methods will likely provide more
  powerful, reproducible, and clinically useful
  diagnoses prognoses
• These tools may enable clinically relevant choice
  of patient and targeted therapy

69
Disclaimer

• I have no financial interest in any of the
  technologies discussed in this lecture other than
  serving as a Scientific Advisory Board member to
  Insert Therapeutics, Inc., for which I receive
  stock and consulting fees.

70
Acknowledgements

• Siwen Hu, MD, PhD
• Daniel Wai, PhD candidate
• Elai Davicioni, PhD candidate
• Deborah Schofield, MD
• Jonathan Buckley, MD, PhD
• Mark Davis, PhD
• Betty Schaub, MT
• COG
• NCI
• Las Madrinas endowment

71
Questions?