Transcript PPT - The International Conference on Bioinformatics InCoB by

Integative Genomic Approaches to Personalized
Cancer Therapy
Patrick Tan, MD PhD
International Conference on Bioinformatics
Singapore, Sept 09 2009
Genomic Oncology in Singapore : Translating
Information into Knowledge
Disease Genes
Clinical Biomarkers
Cancer Pathways
Basic Science to Translation
1) Metastasis Genes
- Network Structures
2) Cancer Classification
- Pathway Signatures
3) Lung Cancer Outcome
- Integrative Genomics
Biological Networks – Robust Yet Fragile
Edge Gene
Hub Gene
Tolerant
Ultrasensitive
Wide Variation
Low Variation
Can we infer ‘hub-like’ genes in cancer?
Yu Kun
Identifying Precisely Controlled Genes in Cancer
Lung
Thyroid
Liver
Esophagus
Breast
270 Tumors
Large
Variation
Restricted
Variation
Restricted Variation Only in Cancers
Cancer
Non-malignant
48 Precisely Controlled Genes in Cancers
The PGC is Precisely Controlled in Many Solid Tumors
Tumor
Significance
Gastric, NPC (99)
Breast (286)
Lung (118)
Ovarian (146)
Breast (189)
Glioma (77)
Colon (100)
0 1 2 3 4 5 6 7 8 9 10 11 12 13
The PGC is NOT Precisely Controlled in Normal Tissues
Normal
Significance
Novartis (158)
Ge et al (36)
0 1 2 3 4 5 6 7 8 9 10 11 12 13
PGC Genes are Enriched in the Integrin Signaling Pathway
Growth Factor Regulation
RAS/MAPK Signaling
PI3K Signaling
JNK/SAPK Signaling
Cytoskeletal Interactions
Cell Motility
Implications of Precise PGC Regulation
Dedicated Cellular Mechanisms to Ensure Accurate
Expression
A Functional Requirement for Tight PGC Control in
Tumors?
Are Tumors Ultrasensitive to PGC Activity?
PGC Expression in Breast Cancer Cell Lines
P=0.01
30 Breast Cancer
Cell Lines
Non-invasive
PGC
Invasive
PGC Expression in Experimental Metastasis
HCT116
Tumor Cells
Splenic
Injection
Liver
Metastases
Adapted from Clark et al (2000)
Reduced PGC Expression Correlates with Metastatic Potential
P=0.022
siRNA Knockdown of PGC Genes Enhances Metastasis
p53CSV siRNA
qRT-PCR
PGC Expression in Primary Tumors
Reduced PGC Expression Predicts Clinical Prognosis
Elevated PGC
Decreased PGC
Are Low-Variance Genes True Hubs?
(Lessons from Yeast)
mRNA variance overlaid on a
protein-protein network
Black nodes = missing data.
A: proteasome regulatory lid
B: mediator complex
C: SAGA complex
D: SWR1 complex
Goel and Wilkins,
unpublished.
Slide Courtesy of Marc Wilkins
Take Home Messages
- A General Strategy for Identifying Tightly Regulated
Genes
- A Precisely Regulated Expression Cassette in Cancer
- Fine-scale alterations potently modulate tumor behaviour
and clinical outcome
-Not discernible by conventional microarray analysis
methods
Yu et al (2008) PLOS Genetics
Basic Science to Translation
1) Metastasis Genes
- Network Structures
2) Cancer Classification
- Pathway Sigantures
3) Lung Cancer Outcome
- Integrative Genomics
High Prevalence of Gastric Cancer in Asia
Global Cancer Mortality
Lung (1.3 million deaths/year)
Stomach (1 million deaths/year)
Liver (662,000 deaths/year)
Colon (655,000 deaths/year)
Breast (502,000 deaths/year)
- WHO, 2005
From The Scientist, Sep 22, 2003
Tumor Heterogeneity Impacts Response
CML
“One Disease”
Gastric Cancer
“Many Diseases”
Imatinib
5-FU
100% Response
20% Response
Pre-Selecting Patients for Optimal Therapy
Gastric Cancer
Subtype A
Rx 1
Subtype B
Rx 2
Subtype C
Subtype D
Subtype E
Rx 3
Rx 4
Rx 5
Subtype F
Rx 6
Expression Signatures as Cancer Phenotypes
Tumor Type B
(“State B”)
Tumor Type A
(“State A”)
Genes
A
B
Expression Signatures
Capture Heterogeneity
Tay et al., Cancer Research (2003)
Using Pathway Signatures to Guide Targeted Therapies
Experimental System
Pathway A
1.5
1
0.5
0
-0.5
-1
-1.5
-1.5
Tumor Profiles
Pathway A
-1
-0.5
0
0.5
1
1.5
Chia Huey Ooi
Mapping Pathway Signatures to Tumor Profiles
Pathway A
-1.5
-1
-0.5
0
0.5
1
1.5
Pathway B
B
C
D
Pathway D
Pathway E
1.5
1.5
1.5
1.5
1.5
1
1
1
1
1
0.5
0.5
0.5
0.5
0.5
0
0
0
0
0
-0.5
-0.5
-0.5
-0.5
-0.5
-1
-1
-1
-1
-1
-1.5
-1.5
-1.5
-1.5
-1.5
-1.5
-1
-0.5
Tumor Profiles
A
Pathway C
0
0.5
1
1.5
-1.5
-1
-0.5
0
0.5
1
1.5
-1.5
-1
-0.5
0
0.5
1
1.5
-1.5
-1
-0.5
0
0.5
1
1.5
Predominant Oncogenic Pathways in Gastric Cancer
200 primary gastric tumors
Oncogenic Pathways
Proliferation
/stem cell
pathways
activated
b-catenin
pathway
activation
p53 pathway
activation
P21
E2F1 (a)
E2F (b)
Stem
0.8 cell (a)
Stem cell (b)
Myc
(a)
0.6
Stem cell (c)
Myc (b)
0.4
NF-kB (a)
Wnt
0.2 (b)
NF-kB
p53 (a)
0
HDAC
b-catenin
Src-0.2
Ras
-0.4
BRCA1
HDAC
p53-0.6
(b)
BRCA1
-0.8
Activation score
-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
-1
0.6
0.8
1
Validating Oncogenic Pathway Predictions
Wnt
Pathways
Proliferation
GC cell lines
NFKB
High Proliferation Scores are Associated with Rapid Growth
Proliferative capacity vs. combined E2F+Myc+Stemcell activation score for
22 GC cell lines
4
Proliferative capacity
Proliferative capacity
3.5
3
2.5
2
1.5
R = 0.5051
p = 0.0165
1
0.2
0.4
0.6
0.8
Summarized activation
of score
the
E2F+Myc+Stemcell
combined score
activation
proliferation/stem cell cluster
1
48h
Linear (48h)
0.6
0.4
0.2
TCF7L2:
SNU16
SNU5
SNU1
KatoIII
NCI-N87
-0.4
YCC3
-0.2
AGS
0
TCF7L2 activity
(folds)
In-silico prediction of
b-catenin pathway activation
High Wnt Scores are Associated with Wnt Activity
Relative constitutive TCF7L2 activity
Oncogenic Pathways in Gastric Cancer are Functionally
Significant
5
p=4.549106
Proliferative
capacity
4
Cell Lines
3
MKN1
2
MKN1
NFKB
0
Control T72/T0
p65 shRNA
shRNA
Annexin +ve cells
60
Wnt
Neg siRNA
b-catenin siRNA
b-catenin (WB)
GC (WB)
cell lines
Actin
% apoptotic
cells
Pathways
1
50
40
30
Cell Death
Assay
20
10
0
NegsiRNA
siRNA
Neg
B-Catenin siRNA
b-catenin
siRNA
Pathway Interactions Influence Survival
Pathway
Combinations
Single
Pathways
NFKB
NFKB +
Prolif.
Proliferation
Wnt +
Wnt
Prolif.
Clinical Validation of Pathway Combinations
Singapore (200)
Proliferation
and NKFB
Proliferation
and Wnt
Australia (90)
Oncogenic Pathways in Gastric Cancer May Guide Therapy
Potential
Therapies
Oncogenic Pathways
200 primary gastric tumors
P21
E2F1 (a)
E2F (b)
Stem
0.8 cell (a)
Stem cell (b)
Myc
(a)
0.6
Stem cell (c)
Myc (b)
0.4
NF-kB (a)
Wnt
0.2 (b)
NF-kB
p53 (a)
0
HDAC
b-catenin
Src-0.2
Ras
-0.4
BRCA1
HDAC
p53-0.6
(b)
BRCA1
-0.8
Activation score
-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
-1
0.6
0.8
1
HLM006474
CX-3543
RTA-402
PXD-101
KX2-391
Salirasib
pifithrin-a
Take Home Messages
•
A framework for mapping defined pathway
signatures into complex tumor profiles
•
Signatures are transportable (in vitro to in vivo)
•
Gastric cancers can be subdivided by pathway
activity into biologically and clinically relevant
subgroups
•
“High-throughput pathway profiling” highlights the
role of oncogenic pathway combinations in clinical
behavior
Ooi et al (2009) PLOS Genetics
Basic Science to Translation
1) Metastasis Genes
- Network Structures
2) Cancer Classification
- Pathway Biology
3) Lung Cancer Outcome
- Integrative Genomics
Genomic Classification of Early Stage
Lung Cancer
Philippe and Sophine Broet
INSERM U472, Faculté de Médecine
Paris-Sud
Lance Miller
Wake Forest University, USA
Broet et al., (2009) Cancer Research
Adjuvant Chemotherapy in Early-Stage NSCLC
Observation
(Watch and Wait)
Surgery
40-50% 5-yr Survival
Chemotherapy?
Stage I, II
Study Questions
Clinical questions
Can we use genomics to discriminate between low
risk (pseudo-stage I) & high risk (pseudo-stage
II) groups?
Previous studies on NSCLC prognosis have been
transcriptome centered, not incorporating
genomic alterations
An Integrated Genomic Strategy to
Identify “Poor Prognosis” NSCLC Cases
Array-CGH
Recurrent Amplifications
And Deletions
Stage IB
NSLCLCs
(Training Set)
Gene Expression Profiling
Highly Regulated
Genes
Recurrent Genomic Alterations in NSCLC
1q31
5p13
CyclinD1
8q24
11q13
WWOX
Genomic Regions Associated with Outcome
Survival associations – “Survival CNAs”
Gene Expression Associated with Survival-CNAs
Gene
Expression
Survival CNAs
Copy Number Driven Expression
203342_at 205564_at 201699_at
202988_at
204322_at
201698_at
203301_at
2113458_at
203343_at
201408_at
Predicting Prognosis in Stage IB NSCLC
Integrated Signature
103 genes (Chr. 7, 16, 20, 22)
Good Prognosis
P=0.002
Poor Prognosis
Training Cohort
Validation of the Integrated Signature
Michigan Series: 73 Stage I A&B NSCLCs
Good Prognosis
P=0.025
Poor Prognosis
Another Validation of the Integrated Signature
Duke Series: 31 Stage I A&B NSCLCs
Good
Prognosis
P=0.003
Poor
Prognosis
Candidates for
Chemotherapy?
Implications for Chemotherapy Selection
Stage II
NSCLC
Poor Prognosis
Stage IB
Poor Prognosis Ib Patients
Are Comparable to Stage II
Patients
Stage Ib
NSCLC
Surgery
Genomic
Predictor
A Genomic Approach
to Guide Chemotherapy
in Early-Stage NSCLC
Good
Prognosis
(“Stage Ia-like”)
Observation
Poor
Prognosis
(“Stage II-like”)
Adjuvant
Chemotherapy
Acknowledgements
Kun Yu
Kumaresan Ganesan
Ooi Chia Huey
Tatiana Ivanova
Shenli Zhang
Wu Yonghui
Lai Ling Cheng
Veena Gopalakrishnan
Jun Hao Koo
Julian Lee
Ming Hui Lee
Iain Tan
Angie Tan
Jiong Tao
Jeanie Wu
Yansong Zhu
Philippe Broet (Paris)
Sophine Broet (Paris)
Lance Miller (GIS)
Elaine Lim (NUH)
Wei Chia Lin (GIS)
Hooi Shing Chuan (NUS)
Alex Boussioutas (Peter Mac, AU)
David Bowtell (Peter Mac, AU)
Sun Yong Rha (S. Korea)
Heike Grabsch (Leeds)
Support :
French-Singapore MERLION program
Singapore Cancer Syndicate
Biomedical Research Council
National Medical Research Council