Transcript cancer - University of Toronto

Cancer Diagnostics
The Old and the New
Course LMP1506S,Thursday,March 7,2002
Eleftherios P. Diamandis, M.D., Ph.D., FRCP(C)
UNIVERSITY OF TORONTO
Laboratory Medicine and Pathology
Compositional analysis of cells, fluids, tissues
(proteins, metabolites, DNA, RNA)
Information invaluable for patient diagnosis,
monitoring, selection of therapy, prognosis,
classification
Timeline of Molecular Pathology
Lakhani and Ashworth Nature Reviews Cancer 2001;1:151-157
Today’s Laboratory Physician / Pathologist
Misconception: Pathologists are those who
perform autopsies and work in isolation by
looking down a microscope all day.
Reality: Participate in teams with surgeons,
oncologists, radiologists; information
provided forms basis for diagnosis and
management and for performing new clinical
trials(by identifying patient groups).
The Current Pathologist and Cancer
Tumor Classification
Essential for cancer prognosis and selection of
treatment:
-Carcinoma vs sarcoma vs lymphoma
-Primary vs metastatic cancer
-Breast carcinoma (ductal vs lobular vs tubular vs mucinous )
-Tumor grade (degree of differentiation)
-Tumor stage (size, lymph node involvement plus imaging
information)
-Surgical margins
The Current Pathologist
Cancer Prognosis
•
Pathologically classified classes of
tumors (by stage, grade, histological
type) behave differently.
•
Different responses to therapy:
ER, PR (+) breast cancers  Tamoxifen
HER2/NEU expression  Herceptin
BCR/ABL translocation  Gleevac
Classical Grading System for Breast Cancer
Classical Staging System for Cancer
The Current Laboratory
Physician / Scientist / Clinical
Pathologist
•
Tumor marker analysis in serum
- screening
- diagnosis
- prognosis
- therapy response
- monitoring for relapse
PSA,CEA,AFP,hCG,CA125,CA15.3
The Problems
Morphology:
•
Subjective analysis - variation between
observers
•
The morphology of the tumor does not
always reveal the underlying biology;
patients with same tumor type can
experience different course of the
disease
•
Immunohistochemistry targets single
molecules; biology depends on many
The Problems
Tumor Markers:
•
No true tumor marker exists (with
notable exceptions)
•
Generally single tumor markers not
good for screening/diagnosis (poor
sensitivity and specificity)
•
Very limited role for predicting
therapeutic response/prognosis
•
Useful as aids for monitoring response
to therapy
Conclusions
We need:
•
better (more objective) and more
biologically-relevant tumor
classification schemes for prognosis,
selection of therapy
•
better tumor markers for population
screening and early diagnosis for cancer
prevention
Paradigm Shift (2000 and Beyond)
Traditional Method: Study one molecule
at a time.
New Method: Multiparametric analysis
(thousands of molecules at a time).
Cancer: Does every cancer have a
unique fingerprint?
(genomic/proteomic?).
The New Laboratory Physician / Scientist /
Pathologist
Changes seen are driven by recent biological / technological
advances:
- Human Genome Project
- Bioinformatics
- Array Analysis
- Mass Spectrometry
_______________________________________
-Automated DNA Sequencing /PCR:
- DNA Arrays
- Protein Arrays
- Tissue Arrays
- Laser Capture Microdissection
- SNPs
- Comparative Genomic Hybridization
Technological
Advances
Microarrays
What is a microarray?
A microarray is a compact device that contains a
large number of well-defined immobilized capture
molecules (e.g. synthetic oligos, PCR products,
proteins, antibodies) assembled in an addressable
format.
You can expose an unknown (test) substance on it
and then examine where the molecule was captured.
You can then derive information on identity and
amount of captured molecule.
AACC 2001
Principles of DNA Microarrays(Printing oligos by
photolithography)
Fodor et al.Science 1991;251:767-773)
Microarray Technology
Manufacture or Purchase Microarray
Hybridize
Detect
Data Analysis
AACC 2001
Applications of Microarrays
• Simultaneous study of gene expression
patterns of genes
• Single nucleotide polymorphism (SNP)
detection
• Sequences by hybridization / genotyping /
mutation detection
• Study protein expression (multianalyte
assay)
• Protein-protein interactions
Provides: Massive parallel information
AACC 2001
Microarray Advantages
• Small volume deposition (nL)
• Minimal wasted reagents
• Access many genes / proteins
simultaneously
• Can be automated
• Quantitative
AACC 2001
If Microarrays Are So Good Why
Didn’t We Use Them Before??
•
•
•
•
Not all genes were available
No SNPs known
No suitable bioinformatics
New proteins now becoming
available
Microarrays and associated technologies should be regarded as
by-products of the Human Genome Initiative and bioinformatics
Limitations of Microarrays
• New technology
• Technical problems (background;reproducibility)
• Need to better define human genes (many
ESTs)
• Manual
• Expensive
AACC 2001
International Genomics Consortium (IGC)
• New initiative
• Aims to generate expression data by
microarrays
• Claims to analyze for all genes 10,000
tumor specimens within 1 year!
• All patients will have detailed follow-up
information
Molecular Signatures/Portraits of
Tumors
Differential Gene Expression
(Budding vs Non-Budding Yeast)
Tissue Expression of KLK6 by Microarray
Highest Expression
brain,spinal cord,then salivary
gland,spleen,kidney
MedianX10
Cell Line or Tissue
Tissue Expression Profiles
• Many proprietary databases - created by
microarray analysis
• Can search as follows:
* which genes are expressed in which tissues
(tissue specific expression)
* unique genes expressed only in one tissue
* quantitative relationships between levels of
expression
* expression is normal vs diseased tissue
Limitations:
- RNA data; not protein
- great variability in results
AACC 2001
Lung Tumor: Up-Regulated
Lung Tumor: Down-Regulated
Whole Genome Biology With Microarrays
Cell cycle in yeast
Study of all yeast genes
simultaneously!
Red;High expression
Blue:Low expression
Lockhart and Winzeler Nature 2000;405:827-836
Microarray Imaging of Tissue Sections
Clinical Care
Diagnosis
Prognosis
Prediction of therapeutic response
Monitoring
Research
Understanding Disease Pathogenesis
Comparative Genomic Hybridization
•
A method of comparing differences in
DNA copy number between tests (e.g.
tumor) and reference samples
•
Can use paraffin-embedded tissues
•
Good method for identifying gene
amplifications or deletions by scanning
the whole genome.
Comparative Genomic Hybridization
Cot1DNA blocks repeats)
Label with Cy-3
Label with Cy-5
Nature Reviews Cancer 2001;1:151-157
Laser Capture Microdissection
An inverted microscope with a low
intensity laser that allows the precise
capture of single or defined cell groups
from frozen or paraffin-embedded
histological sections
Allows working with well-defined
clinical material.
Tumor Heterogeneity(Prostate Cancer)
Tumor Cells, Red
Rubin MA J Pathol 2001;195;80-86
Benign Glands,Blue
Laser Capture Microdissection
LCM uses a laser beam and a special thermoplastic polymer transfer cup(A).The cap is set on the surface
of the tissue and a laser pulse is sent through the transparent cap,expanding the thermoplastic polymer.
The selected cells are now adherent to the transfer cap and can be lifted off the tissue and placed directly
onto an eppendorf tube for extraction(B).
Rubin MA,J Pathol 2001;195:80-86
Tissue Microarrays
• Printing on a slide tiny amounts of tissue
• Array many patients in one slide (e.g. 500)
• Process all at once (e.g. immunohistochemistry)
• Works with archival tissue (paraffin
blocks)
AACC 2001
Gene Expression Analysis of Tumors
cDNA Microarray
Lakhani and Ashworth Nature Reviews Cancer 2001;1:151-157
Tissue Microarray
Alizadeh et al J Pathol 2001;195:41-52
Histochemical staining of microarray tissue cores of
ovarian serous adenocarcinoma. -tjc Identical microscopic fields showing variable staining
intensity of various tissue cores for HK6 (right)
• H&E
• HK6
Histochemical staining of a microarray tissue core of ovarian
clear cell adenocarcinoma. -tjcIdentical microscopic fields showing strong cytoplasmic positivity
for HK6 within carcinoma (and endothelium, lower right)
• H&E
• HK6
Histochemical staining of a microarray tissue
core of ovarian serous adenocarcinoma. -tjcNote: Cytoplasmic positivity for HK6 in carcinoma,
endothelium and stromal cells.
• H&E
• HK6
Molecular Profiling of Prostate Cancer
Rubin MA,
J Pathol 2001;195:80-86
Single Nucleotide Polymorphisms (SNP)
• DNA variation at one base pair level; found at a
frequency of 1 SNP per 1,000 - 2,000 bases
• Currently, a map of 1.42 x 106 SNPs have been
described in humans (Nature 2001; 409:928-933)
by the International SNP map working group)
• Identification: Mainly a by-product of human
genome sequencing at a depth of x10 and
overlapping clones
• 60,000 SNPs fall within exons; the rest are in
introns
AACC 2001
Why Are SNPs Useful?
• Human genetic diversity depends on SNPs
between individuals (these are our genetic
differences!)
• Specific combinations of alleles (called “The
Haplotype”) seem to play a major role in our
genetic diversity
• How does this genotype affect the phenotype
Disease
predisposition?
Continued:……..
Why are SNPs useful………………..continued:
Diagnostic Application
Determine somebody’s haplotype (sets of
SNPs) and assess disease risk.
Be careful: These disease-related haplotypes
are not as yet known!
AACC 2001
Genotyping: SNP Microarray
 Immobilized allele specific oligo probes
 Hybridize with labeled PCR product
 Assay multiple SNPs on a single array
TTAGCTAGTCTGGACATTAGCCATGCGGAT
GACCTGTAATCG
TTAGCTAGTCTGGACATTAGCCATGCGGAT
GACCTATAATCG
High- Throughput Proteomic Analysis
By Mass Spectrometry
Haab et al Genome Biology 2000;1:1-22
Applications of Protein
Microarrays
 Screening for Small molecule
targets
Post-translational
modifications
 Protein-protein
interactions
Protein-DNA
interactions
 Enzyme assays
 Epitope mapping
Cytokine Specific Microarray ELISA
IL-1 
IL-6
IL-10
marker protein
cytokine
Detection system
BIOTINYLATED MAB
ANTIGEN
CAPTURE MAB
VEGF
MIX
Recently
Published
Examples
Rationale For Improved Subclassification
of Cancer by Microarray Analysis
•
Classically classified tumors are
clinically very heterogeneous - some
respond very well to chemotherapy;
some do not.
Hypothesis
The phenotypic diversity of cancer might be
accompanied by a corresponding diversity in
gene expression patterns that can be captured
by using cDNA microarrays
Then
Systematic investigation of gene expression
patterns in human tumors might provide the
basis of an improved taxonomy of cancer.

Molecular portraits of cancer
Molecular signatures
Molecular Portraits of Cancer
Breast Cancer
Perou et al Nature 2000;406:747-752
Green:Gene underexpression
Black:Equal Expression
Red:Overexpression
Left Panel:Cell Lines
Right Panel:Breast Tumors
Figure Represents 1753 Genes
Differential Diagnosis of
Childhood Malignancies
Ewing Sarcoma:Yellow
Rhabdomyosarcoma:Red
Burkitt Lymphoma:Blue
Neuroblastoma:green
Khan et al.Nature Medicine 2001;7:673-679
Differential Diagnosis of Childhood Malignancies
(small round blue-cell tumors,SRBCT)
EWS=Ewing Sarcoma
NB=Neuroblastoma
RMS=Rhabdomyosarcoma
BL=Burkitt Lymphoma
Note the relatively small number of
genes necessary for complete
discrimination
Khan et al.Nature Medicine 2001;7:673-679
Aggressive vs Non-Aggressive Breast Cancer Cell Lines
Can accurately predict
aggressiveness with a
set of only 24 genes
Zajchowski et al
Cancer Res 2001;61:5168-78
Selected Applications of Microarrays
Alizadeh et al. Nature 2000;403:503-511
•
Identified two very distinct forms of
large B-cell Lymphoma
•
The two forms had different clinical
outcomes (overall survival).
Conclusion
Molecular classification of tumors on the
basis of gene expression can identify
previously undetected and clinically
significant subtypes of cancer.
Novel Classification of Lymphoma
Alizadeh et al
Nature 2000;403:503-511
GI Tumors with KIT Mutations
A:IHC with KIT antibody(negative)
B:IHC with KIT antibody(positive)
C:Multidimensional scaling plot
Orange Dots:KIT mutation-positive
Gastrointestinal Stromal Tumors
Blue Dots:Spindle Cell Carcinomas
Allander et al.Cancer Res 2001;61:8624-8628
Gene Expression Profile of GI Stromal Tumors with KIT Mutations
KIT (-)
KIT(+)
KIT gene
Allander et al.Cancer Res 2001;61:8624-8628
Applications (continued)
Vant’t Veer L. et al. Nature 2002:415-586
Examine lymph node negative breast cancer patients and
identified specific signatures for:
* Poor prognosis
* BRCA carriers
The “poor prognosis” signature consisted of genes regulating
cell cycle invasion, metastasis and angiogenesis.
Conclusion
• This gene expression profile will outperform all currentlyused clinical parameters in predicting disease outcome
• This may be a good strategy to select node-negative
patients who would benefit from adjuvant therapy.
Prognostic Signature of Breast Cancer
Patients above line
No Distant Metastasis
Patients Below Line
Distant metastasis
Van’t Veer et al.Nature 2002;415:530-536
ER(+)vs ER(-) Signatures in Breast Cancer
Sporadic vs BRCA1 Signatures in Breast Cancer
Patients above line:ER(+)
Patients below line:ER(-)
Patients above line:BRCA1-positive
Patients below line:BRCA1-negative
Van’t Veer et al.Nature 2002;415:530-536
Molecular Signatures for Selecting Treatment Options
Van’t Veer et al.Nature 2002;415:530-536
Strategies
to Discover New Cancer
Markers
Human Genome
Establish tissue expression of all human genes by
microarray technology
Identify “tissue-specific” genes
Compare “normal” vs “cancer”
Select highly overexpressed genes
Evaluate in detail
Many potential pitfalls
An Example of Genome Mining Approach
to Discovery Circulating Markers for
Ovarian Carcinoma.
Welsh JB, et al., PNAS 2001; 98: 1176 - 1181
AACC 2001
Method
49 arrays on a 7x7 Matrix
ARRAYS ON ARRAYS
Hybridize 49 different
samples in one shot (some
normal; some malignant)
6,000 genes per array
A.Tumor Classification
B.Expression of genes
RED:Overexpression
GREEN:Underexpression
Genes Overexpressed in Ovarian
Cancer
From:Welsh et al PNAS 2001;98:1176-1181
Genes Overexpressed in Ovarian
Cancer(RT-PCR Verification)
CD 24
HE4
LU
rRNA
From:Welsh et al PNAS 2001:98:1176-1181
Mass Spectrometry for Proteomic
Pattern Generation
•
Serum analysis by SELDI-TOF
mass spectrometry after extraction
of lower molecular weight proteins
•
Data analyzed by a “pattern
recognition” algorithm
Serum Fingerprint by Mass Spectrometry
Results
_________________________________________________
Classification by Proteomic Pattern
Cancer Unaffected New Cluster
________________________________________________
Unaffected Women
No evidence of ovarian cysts
2/24
22/24
0/24
Benign ovarian cysts <2.5cm
1/19
18/19
0/19
Benign ovarian cysts >2.5cm
0/6
6/6
0/6
Benign gynecological
0/7
0/7
7/7
inflammatory disorder
__________________________________________________________
Women with Ovarian Cancer
Stage I
18/18
0/18
0/18
Stage II, III, IV
32/32
0/32
0/32
__________________________________________________________
Petricoin III EF, et al. Lancet 2002;359:572-577
The Future??
Cancer Patient

Surgery/Biopsy
Cancerous Tissue

Array Analysis
Tumor Fingerprint

Individualized Treatment
The Future??
General Population
- Imaging
- Multiparametric/
miniature testing of
serum on a protein array
- Mass spectrometric
serum/urine proteomic
pattern generation
Screen-positive patients
Prevention; Effective Therapy
The Future?
Asymptomatic individuals
Whole genome SNP
analysis
Predisposition to certain disease
Prevention (drugs; lifestyle)
Surveillance
The Future?
•
•
•
•
•
•
•
Miniature ingestible or intravenous diagnostic devices
will provide “images” or “information” related to
body function.
Wristwatch devices for biomonitoring
Telemedicine-Videoconferencing
Electronic medical record
New, highly effective therapies
“Electronic” behavioural modification
Gene therapy.
NONE OF THE ABOVE
Young and Wilson, Clin Cancer Res 2002;8:11-16