Microarray analysis as a

1
Microarray analysis as a prognostic and
predictive tool are we ready? Enzo
MedicoLaboratory of Functional Oncogenomics
Institute for Cancer Research and
TreatmentUniversity of Torinoenzo.medico_at_ircc.it
2
Topics

• Platforms for gene expression profiling
• Breast cancer signatures
• From cell-based models to cancer classifiers

3
AFFYMETRIX GeneChip
4
AFFYMETRIX GeneChip
45,000 gènes !
5
The Probe Sets
6
Hybridization on the chip
7
Signal detection
8
Genomic raw data
9
Gene expression profilingby spotted microarrays
Oligonucleotides or cDNAs
Robotic printing
10
Gene expression profilingby spotted/dual colour
microarrays
RNA extraction, cDNA labelling
Hybridization
11
Different platforms generatedifferent data types
Two-colour
One-colour
Paired samples
Independent samples
12
Topics

• Platforms for gene expression profiling
• Breast cancer signatures
• From cell-based models to cancer classifiers

13
SHOULD ONE TREAT A SMALL (lt1CM) ENDOCRINE
UNRESPONSIVE TUMOR ?

Choices of 40 experts worldwide
48
61 years IDC Postmenopausal N - pT 0.9 cm
Grade 2 ER et PgR - HER2 -
25
8
15
4
FA(E)C x 6
14
THERAPY DECISION-MAKING FOR EARLY BREAST CANCER

WHO CAN BE SPARED THERAPY?
WHICH THERAPY WILL WORK BEST?
Prognostic factors needed
Predictive factors needed
15
(No Transcript)
16
The Intrinsic Breast Cancer Signatures
17
Confirmatory Study
ER-
ER
PNAS vol 100, no 18, 10393-10398, 2003
Clinical Outcome
18
Discovery of poor prognosis signatures for
distant relapses
Amsterdams Signature 312 patients 70 genes
Rotterdams Signature 286 patients 76 genes
19
70-gene poor prognosis signature
78 tumor samples
20
70-gene expression signature outperforms
clinicopathological criteria
High Risk
Low Risk
Marc J Van de Vijver et al., NEJM, 347, 25, 2002
21
286 tumor samples
Lancet, 2005, 365, 671-679
22
Sotiriou et al., JNCI 2006

• Poor inter observer reproducibility
• G2 difficult treatment decision making, under-
  or over-treatment likely

• Findings consistent across multiple data sets
  and microarray platforms
• More objective assessment
• Easier treatment decision-making
• High proportion of genes involved in cell
  proliferation !

23
Definition and validation of the Genomic Grade
Analyze on validation set (n 125)?
Identify genes correlated with grade 1 vs grade 3
Grade 1
Grade 2
Grade 3
Grade 1
Grade 3
24
Consistent Distribution of GG in Different
Populations and Microarrays Platforms
Sorlie et al. PNAS 2001
Van de Vijver et al. NEJM 2002 Central Pathology
Review!
Sotiriou et al. PNAS 2003
25
GENE EXPRESSION SIGNATURE POWERFUL PROGNOSTIC
TOOL
Highest priority Transfer from bench to bedside
HOW ?
26
Validation study
27
(No Transcript)
28
THERAPY DECISION-MAKING FOR EARLY BREAST CANCER
WHO CAN BE SPARED THERAPY?
WHICH THERAPY WILL WORK BEST?
Prognostic factors needed
Predictive factors needed
29
(No Transcript)
30
Topics

• Platforms for gene expression profiling
• Breast cancer signatures
• From cell-based models to cancer classifiers

31
The Invasive Growth biological program
Scattering and migration
Differentiation, cell polarity, tubulogenesis
Proliferation
Survival and protection against apoptosis
32
MLP-29 liver stem/progenitor cells activate the
invasive growth program in response to HGF
CTRL
Day 1

• MET
• SHH
• SMO
• PTCH1
• GLI
• AFP
• CK19
• ALB
• ?-AT
• TO

Hedgehog pathway
HGF 6h
Day 2
Liver lineage
HGF 16h
Day 4
Liver differentiation
-3
3
MLP29 / liver log2 ratio
33
The Invasive Growth Transcriptional Program
HGF/CTRL 1h 6h 24h
EGF/CTRL 1h 6h 24h
HGF/CTRL 1h 6h 24h
EGF/CTRL 1h 6h 24h
HGF/CTRL 1h 6h 24h
EGF/CTRL 1h 6h 24h
1
Induced at 1h
10
6
2
Suppr. at 1h
7
3
11
8
Induced at 6h
Suppressed at 6h
4
9
12
Suppressed at 24h
10
13
Induced at 24h
5
11
14
15
12
Suppressed at 24h
13
14
15
34
Classifier construction and in silico validation
using breast cancer microarray datasets
Total NKI Breast cancer Dataset (311 samples -
Agilent)?
Rotterdam Breast cancer Dataset (286 samples -
Affymetrix)?
IG genes ranked by their individual
performance (SNR over 1000 bootstraps)?
Statistical analysis
Number of genes in the classifier optimized and
definition of the nearest mean classifier (NMC)?
Kaplan-Meier
COX proportional hazard
35
The Nearest Mean Classifier
Training Group A
Training Group B
36
The Nearest Mean Classifier
Group A
Group B
Pearson correlation -gt classification
37
Invasive growth genes classify breast
cancersamples by their metastatic propensity
38
Validation on the Rotterdam dataset(286 breast
samples, Wang et all., Lancet, 2005)?
Legend 0 Good prognosis samples 1 Poor
prognosis samples
39
Breast cancer expression profiling towards an
integrated approach to personalized therapy
40
Acknowledgments
IRCC Laboratory of Functional Oncogenomics Tomma
so Renzulli Claudio Isella Daniela
Cantarella Barbara Martinoglio Roberta
Porporato IRCC Gynaecological Oncology Daniela
Cimino Luca Fuso Prof. Michele De Bortoli Prof.
Piero Sismondi