Download Cancer Presentation Final 101602

Biomarker Discovery
台大醫學院生化暨分生所
周綠蘋
Predicting Life Processes
(storage)
DNA
Transcription
Gene Expression
Translation
Proteins
Environment
Proteomics
Biochemical Circuitry
Metabolomics
Phenotypes (Traits)
High Throughput Technologies

Genetic

Microarray

Proteomic
Why Proteomics?
Same Genome
Different Proteome
Biobarkers
Biomarkers can play roles before cancer diagnosis (in risk assessment
and screening), at diagnosis and after diagnosis (in monitoring therapy,
selecting additional therapy and detecting recurrence).
The BRCA1 (breast cancer 1) gene can be used in breast cancer, both for
risk assessment and as a predictor of 10-year survival.
Serum CEA is increased in colon, breast and lung cancer, but also in
many benign conditions.
AFP and b-HCG can be increased for many reasons, their reliability in
assessing testicular cancer burden following diagnosis accounts for their
integration into staging.
PSA, cancer antigen 125, CA19-9, and other, similar markers often fails
to correlate with tumour burden.
Clinical Proteomics
Answering Clinical Questions
with Biomarker Patterns
Clinical Applications of ProteinChip Technology


Find candidate disease markers that can be used for prediction,
diagnosis and/or monitoring of therapy.
Dissect pathogenesis and mechanisms
of disease.

Identify markers that predict response to
potential drugs (clinical stratification and
outcome prediction).

Identify markers that can be used to
monitor drug toxicity.
Limitations of Current Cancer Diagnostics




Most of current diagnostics detect cancer after it has already spread
to other parts of the body
Currently single biomarkers are used for cancer diagnosis which
have lower sensitivity and specificity
Early detection of cancer improves treatment options and survival
rates
Novel high value multi-marker diagnostics can be discovered by
profiling of serum and tissue samples
ProteinChip® Technology
ProteinChip Arrays and SELDI-TOF-MS Detection
1. Sample
2. Proteins
goes directly onto the ProteinChip Array
are captured, retained and purified directly on the chip (affinity capture
)
3. Surface is “read” by Surface-Enhanced Laser Desorption/Ionization (SELDI)
4. Retained proteins can be processed directly on the chip
Laser
Molecular Weight
Sample
100 m2
to
1 mm2
ProteinChip Array
Chemical, Biochemical or Biological Capture Surface
ProteinChip® Array
Technology
Chemical Surfaces – Protein Expression Profiling:
Hydrophobic
Anionic
Metal Ion
Cationic
Normal Phase
Biological Surfaces – Protein Interaction Assays:
PS-1 or PS-2
Antibody - Antigen
Receptor - Ligand
DNA - Protein
ProteinChip® Array Preparation
1. Apply Crude Sample
Proteins bind to chemical or biological
“docking sites” on the ProteinChip Array
surface through an affinity interaction
2. Wash ProteinChip Array
Proteins that bind non-specifically and
buffer contaminants are washed away,
eliminating sample “noise”
3. Add Energy Absorbing Molecules or “Matrix”
After sample processing the Array is dried and EAM is
applied to each spot to facilitate desorption and ionization
in the TOF-MS
Detection Using the ProteinChip® System

Retained proteins are “eluted” from the ProteinChip Array by Laser
Desorption and Ionization
Ionized proteins are detected and their mass accurately determined
by Time-of-Flight Mass Spectrometry
Detector

TOF-MS
5000
Trace View
7500
10000
12500
10000
10000
12500
12500
10
5
5000
10
5000
Gel View
5
7500
0
7500
Map View
10000
5000
5000
12500
7500
7500
Weak cation exchange pH
4.5 wash
10000
12500
0
5000
5000
5000
5000
7500
7500
10000
10000
7500
7500
12500
12500
Weak cation exchange pH
10000
10000
4.5 wash
12500
12500
ProteinChip® Systems and Arrays
血清蛋白質指纹圖譜
Cancer Biomarker Discovery
2
9781.51
12500
Ret Map 1-4
Patient
9234.28
9975.5
9227.03
8711.41
8906
10000
8098.46
9404.62
7912.93
7516.78
7500
7699.22
4
5000
6229.73
6422.12
6615.94
2500
3028.2+H
6
4469.89
0
3994.89
2
8908.66
7500
7700.05
7907.11
8092.4+H
4
6226.76
6419.11
6605.27
6
5000
4843.6+H
2500
4467.6+H
3997.86
A Real Example
10000
12500
Control
Ret Map 1-3
0
2500
5000
7500
10000
12500
Ret Map 1-4(2)
GelViewTM
Ret Map 1-3(2)
7.5
5
Difference
map
comp1
2.5
0
-2.5
2500
2500
5000
5000
7500
7500
10000
10000
12500
12500
6
4
2
0
2500
5000
2500
2500
5000
5000
5000
2
0
2500
4469.89
5000
7500
5000
5000
7500
7500
7500
7500
7500
7500
9404.62
9234.28
9975.5
8711.41
9781.51
9234.28
9227.03
8906
8908.66
9404.62
8908.66
8098.46
7500
7700.05
7907.11
8092.4+H
7500
7912.93
3997.86
6226.76
6419.11
6605.27
4843.6+H
4467.6+H
9975.5
9781.
9227.03
8906
8711.41
7700.05
7912.93
7907.11
8092.4+H
6226.76
6419.11
7516.78
6605.27
8098.46
7516.78
7699.22
5000
6229.73
6422.12
6615.94
5000
7500
7699.22
6
4467.6+H
6229.73
6422.12
4843.6+H
6615.94
3997.86
2500
3994.89
4
3028.2+H
4469.89
3994.89
A real example
in cancer
5000
10000
10000
10000
10000
10000
10000
2.5
10000
10000
10000
10000
12500
R
12500
Patient
Ret Map 1-4
12500
12500
12500
Control
R
Ret Map 1-3
12500
Ret Map 1-4(2)
7.5
-2.5
12500
12500
12500
12500
R
GelViewTM
Ret Map 1-3(2)
R
5
comp1
Difference
map
0
c
The Impact of Multiple Biomarkers &
Biomarker Patterns Software
Peak A
Criteria
Peak B
Criteria
Cancer
Normal
Peak C
Criteria
Cancer
ID the biomarkers,
Link to biology of
disease
Normal
Sensitivity
Specificity
(“True
Positives”)
(“True
Negatives”)
Single Marker
65%
35%
Biomarker Pattern
97%
83%
Faster Biomarker Discovery
20.0
5000
7500
10000
12500
15000
15.0
100
75
50
25
0
Control
10.0
5.0
100
75
50
25
0
Control
100
75
50
25
0
Control
100
75
50
25
0
Treated
Control
Groups
Treated
20.0
15.0
10.0
10
Control
5.0
0
100
75
50
25
0
10
100
75
50
25
0
10
Control
0
Treated
10
Control
Groups
Treated
Control
0
Treated
0
Treated
10
Treated
0
5000
7500
10000
12500
10
Treated
15000
0
11000
11500
12000
12500
13000
Under specific bind/wash
conditions, look for statistically
significant up and down
regulation of protein signals
ProteinChip® Array Application:
Breast Cancer Biomarker Discovery
Background





Breast cancer is the most common form of cancer (other than skin
cancer) in women
200,000 new cases of breast cancer detected each year of which
40,000 will die
Although mammography increased awareness, its effectiveness is
still being investigated
CA15.3, a serum biomarker is being is being tested for use in
breast cancer detection but it has low sensitivity (23%) and
specificity (69%)
Multiple markers with higher specificity and sensitivity can
improve early detection of breast cancer & save lives
Clinical Design
103 Breast cancer sera
4
Stage 0
38
Stage I
37
Stage II
24
Stage III
66 Non-cancer control sera
25
Benign breast disease
41
Healthy Control
Representative Spectra of Selected Biomarkers
Distribution of BC3 across all diagnostic
groups
1.6
BC3 (8.9KD)
1.2
0.8
0.4
0
-0.4
Healthy Donor
100%
Benign
76%
Specificity = 91%
Stages 0-I
88%
Stage II
78%
Stage III
92%
Sensitivity = 85%
Distribution of the composite index across all
diagnostic groups
LR Composite Index
1.2
0.6
0
-0.6
-1.2
Healthy Donor
100%
Benign
85%
Specificity = 91%
Stages 0-I
93%
Stage II
85%
Stage III
94%
Sensitivity = 93%
Improved cancer diagnostic using
multiple SELDI markers
1
+
Sensitivity
0.8
0.6
+
0.4
Composite Index (AUC=0.972)
BC1 (AUC=0.846, p<0.0001)
BC2 (AUC=0.795, p<0.0001)
BC3 (AUC=0.934, p<0.01)
p-values: AUC comparison of indiv.
biomarker against composite index
0.2
0
0
0.1
0.2
0.3
0.4
0.5
0.6
1-Specificity
0.7
0.8
0.9
1
Breast Cancer Biomarker Study
Conclusions
Specificity (True Negative Ratio)
Sensitivity (True Positive Ratio)
Single Marker
CA15.3
Multiple Markers (BC1-3)
by SELDI Profiling
69%
91%
23%
93%
Improving Clinical Confidence
Multi-marker analysis


Current applications of single marker assays

Confirmation of diagnosis

Limited monitoring
Potential applications of multi-marker assays

Early detection

Correct diagnosis

Staging/severity assessment

Treatment targeting

Prognosis

Real-time monitoring of treatment response


Clinical trial stratification to aid assessment of efficacy
and side-effects
Sensitive, full spectrum, toxicology assessment
SELDI血清蛋白質指纹圖譜-癌症的早期診斷
0
5000
10000
15000
N
0
5000
10000
15000
0
5000
10000
15000
0
5000
10000
15000
0
5000
10000
15000
0
5000
10000
15000
0
5000
10000
15000
0
5000
10000
15000
0
5000
10000
15000
00
5000
5000
10000
10000
15000
15000
A
B
C
D
E
F
G
H
A. 肝癌
B. 肺癌
C. 胃癌
D. 腸癌
E. 卵巢癌
F. 乳腺癌
G. 胰腺癌
H. 前列腺癌
Cancer Markers

肝癌: 甲胎蛋白 AFP
(靈敏度 42%)
SELDI
(靈敏度 91%, 特異性 89%)

肺癌: NSE
(靈敏度 <50%)
SELDI
(靈敏度 82%, 特異性 95%)

胃癌: 癌胚抗原 CEA
(靈敏度 <50%)
SELDI
(靈敏度 91%, 特異性 94%)

前列腺癌: PSA (靈敏度 <50%)
SELDI (靈敏度 83%, 特異性 97%)

乳腺癌: 癌抗原 CA153
(靈敏度 23%)
SELDI
(靈敏度 93%, 特異性 91%)

卵巢癌: 癌抗原 CA125
(靈敏度 34%)
SELDI
(靈敏度 99%, 特異性 99%)

腸癌: 癌胚抗原 CEA
(靈敏度 47%)
SELDI
(靈敏度 83%, 特異性 92%)

胰腺癌: CA199 (靈敏度 <50%)
SELDI (靈敏度 85%, 特異性 82%)

Bladder (urine cytology )(靈敏度 40%)
SELDI
(靈敏度 80%, 特異性 86%)

NPC
SELDI
(靈敏度 92%, 特異性 97%)
Protein Based Drug Discovery
Therapeutic Drugs
Inhibitors of HIV Replication
ADARC (David Ho, Liqui Zhang and colleagues)
Ciphergen Biomarker Discovery Center
Science 2002
HIV Infects Cells That Express
CD4 Receptors
CD4 receptors are found in
T-cells, macrophages and
dendritic cells
Two co-receptros of HIV are
Known as CCR5 & CXR4
R5 strains of HIV bind CCR5
Found on the surface of
Macrophages & dendritic cells
X4 strains of HIV bind CXR4
Found on T-cells
CAF suppresses viral production by infected cells
but does not kill the infected CD-4+ cell
CD4 cell
Viral peptide antigens
expressed on cell
surface in MHC-I
molecules
CD4 cell
CAF secreted by CD8+ T cells
HIV Infects
CD4+ helper Tcell
CD4 cell
CD8+ cells
suppress HIV
production
Characteristics of
CD8+ Anti-viral Factor (CAF)

Present in long term non-progressors (LTNP)

Not usually detected in normally progressing AIDS patients

Stable at high temperatures (560 C for 30 min or 1000 C for 10 min)

Stable at low pH (conditions ranging from pH 2-8)

Low MW (under 10 KD)

Resistant to trypsin; sensitive to Staphylococcus V8 protease

Lacks identity to other known cytokines

Inhibits HIV replication regardless of viral phenotype or tropism
Discovery of a cluster of small proteins
secreted by stimulated CD8+ T cells: LTNP-3
LTNP-3 stimulated
3371.9
*
*
7815.0
3442.5
LTNP-3 unstimulated
*
EC-SELDI-MS Platform Protein Identification
Using ProteinChip Interface
Protein Sequencing by Collision Induced Dissociation
Directly from the ProteinChip Surface
Collision
Cell
ProteinChip
Ion Guide
Scanning
Quadrupole
Pulser
Detector
Microchannel
Plates
UV Beam
Ion Mirror
Tandem Mass Spectrometry - MS/MS





Ions are produced from
all peptides in the
sample in ion source
Only one type of
peptide passes through
MS1
This peptide collides
with gas molecules in
Q2 and fragments
The fragments from
this peptide are
analyzed in MS2
This MS/MS spectrum
provides the amino
acid sequence of this
peptide
MS1
MASS FILTER
Q2
MS2
Collision Cell MASS FILTER
ION
DETECTOR
ION
SOURCE
PRECURSOR ION
SEPARATION
MS
Spectrum
NEUTRAL GAS
COLLISIONS
PRODUCT ION
SEPARATION
MS/MS
Spectrum
MS/MS Spectrum of Peptides
- amino acid sequencing from peptide fragments
zn-1
zn-2
yn-1
xn-1
z2
z1
yn-2
+2H
xn-2
y1
+2H
x1
+2H
NH2_ CH_ CO_ NH_ CH_ CO_ NH_ CH_ CO……...... NH_ CH_ CO_ NH_ CH_ CO2H
R1
R2
a1
R3
Rn-1
a2
b1 +2H
c1
Rn
an-1
b2
+2H
c2
+2H
bn-1 +2H
cn-2
cn-1
- y and b ions predominate in low energy MS/MS of tryptic peptides
- mass difference between ions of same type (i.e. - y1 and y2)
correspond to amino acid molecular weight
CID of tryptic digest fragment 1216.63
Copyrighted by Science, 2002
MS-Tag Search Results of 1216.6 tryptic
fragment CID
The peptide fragment is identical to a
region of human a-defensin-1, -2 and -3
10
20
30
Human a-defensin-1 ACYCRIPACIAGERRYGTCIYQGRLWAICC
Human a-defensin-2 -CYCRIPACIAGERRYGTCIYQGRLWAICC
Human a-defensin-3 DCYCRIPACIAGERRYGTCIYQGRLWAICC
What is known about human a -defensins?
 Antibiotic peptides produced by
neutrophils
 They display cationic properties
Hill et al., Science 1991
Multidimentional Protein Profiling
1. Surface type
SO3
SO3
SO3
NR3
NR3
NR3
NR3
Me(II) Me(II)
Me(II)
SO3
0.15 M NaCl
1.0 M NaCl
Buffer pH 7
Buffer pH 9
3 M urea
6 M urea
0.1% Tween 20
0.1% CHAPS
7.5
3. Native Mass
5
2.5
0
2000
4000
6000
8000
A standard experiment to characterize a protein or protein mixture would be to choose a set
of array surface types and washing conditions. Depending on the chosen surface and/or
conditions only a subset of proteins in the complex biological sample’s protein mixture will
bind in a specific manner. Which proteins are bound is analysed in the spectra obtained from
the MS readout.
PatternTrack® Process
Cancer Research 64, 5882–5890, August 15, 2004
Three Biomarkers Identified from
Serum Proteomic Analysis for the
Detection of Early Stage Ovarian Cancer
Zhen Zhang, Robert C. Bast, Jr., Yinhua Yu, Jinong Li, Lori J. Sokoll, Alex J. Rai, Jason M.
Rosenzweig, Bonnie Cameron, Young Y. Wang, Xiao-Ying Meng, Andrew Berchuck, Carolien van
Haaften-Day, Neville F. Hacker, Henk W. A. de Bruijn, Ate G. J. van der Zee, Ian J. Jacobs, Eric T.
Fung, and Daniel W. Chan
Use of SELDI for discovery, validation, and identification of
biomarkers for the detection of early stage ovarian cancer
Study Design

Discovery
503 serum samples were collected at four medical institutions for discovery phase.


153 patients with invasive epithelial ovarian cancer

42 with other ovarian cancers

166 with benign pelvic masses

142 healthy women
Validation and Multi-Variate Pattern Analysis



A multi-variate pattern including three biomarkers and CA-125 was determined for diagnosis
of early stage (I/II) ovarian cancer.
Purification and Identification


Cross validation with samples from different medical institutions was performed to ensure
biomarker association with ovarian cancer.
Three biomarkers were purified and further identified using SELDI technology.
Assay

142 independently archived samples were collected at a fifth medical institute. Samples
were tested using an independent immunoassay to validate the biomarker panel.
Multi-Center Study Design
Groningen Univ Hosp
Ca I/II (20)
Discovery
Set 2
H. Control (30)
Benign (50)
Duke Univ Med Cntr
Ca I/II (35)
Discovery
Set 1
Selected Biomarkers
and CA125
H. Control (49)
Royal Hosp for Women
Ca I/II (35)
Test Set
Ca: 29, HC: 46
Multivariate
Model
Derivation
Ca III/IV (2)
Benign (90)
Training Set
Ca: 28, HC: 33
Cross-Validation
Multivariate
Predictive Models
Candidate Biomarkers
Ca III/IV (103)
Benign (26)
MD Anderson Ca Cntr
H. Control (63)
Independent
Validation
Validation Set 1
Johns Hopkins Med Inst
Ca III/IV (41)
Breast Ca (20)
Colon Ca (20)
Prost. Ca (20)
H. Control (41)
Immunoassay Test
Validation Set 2
Protein
Identification
Selected Biomarkers
and CA125
Identified
Biomarkers
Independent
Validation
Validation Set 1
Automated Biomarker Discovery
using Pattern Track™ Technology
Copper Ion
Anion Exchange
Hydrophobic
pH9+FT
pH7
Denatured
Serum
Sample
pH5
pH4
Replicate 1
Replicate 2
Replicate 3
Cation Exchange
Randomized profiling of
replicates on 4 arrays
using Biomek® 2000 robot
pH3
organic
Automated pre-fractionation
using 96-well microplates
containing anion exchange
chromatographic sorbent
ProteinChip
System SELDI
TOF-MS
Calibrate
Baseline Subtract
Normalize
25000
50000
75000
25000
50000
75000
20
15
10
5
0
Multi-variate Analysis
SELDI Analysis of Fractionated Serum from
Ovarian Cancer Patients and Healthy Women
Fraction pH4, IMAC-Cu
Fraction pH9, IMAC-Cu
Stage I ovarian
cancer patient 1
Stage I ovarian
cancer patient 2
Healthy
woman 1
Healthy
woman 2
m/z 12,828
m/z 28,043
m/z 3,272
ProteinChip Assisted Protein
Purification Strategy
NP20 ProteinChip Arrays
Identification of Three Biomarkers
Differentially Expressed Peaks
3,272 Da
Up-regulated in ovarian cancer samples
12,828 Da
Down-regulated in ovarian cancer samples
28,043 Da
Down-regulated in ovarian cancer samples
Biomarker Identity
Fragment of inter-a-trypsin
inhibitor, heavy chain H4
Truncated form of transthyretin
Apolipoprotein A1
Model Comparison Using Receiver
Operating Characteristic (ROC) Curves
1.0
0.9
0.8
Sensitivity
0.7
0.6
0.5
— CA125, AUC=0.770
0.4
— 3 Markers + CA125, AUC=0.885, p=0.023
— 3 Markers, AUC=0.920, p=0.028
0.3
0.2
0.1
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
1 - Specificity
Independent validation set: stages I/II epithelial ovarian cancer vs. healthy controls.
OvaRI assay clinical trial
Diagnostic Markers for Tuberculosis
Lancet 2006; 368: 1012–21
Serum
179
TB
170
Controls
A key advantage of SELDI-ToF MS
lies in the discovery phase, which can profile large
numbers of samples in a high throughput fashion, and
by using whole signatures, reduce problems with
individual variability in peak detection.
CM10 ProteinChip for protein profiling
Multiple
Biomarkers
ProteinChip® Biomarker Discovery –
Multi-marker Discovery, Validation & Screening
OUTPUT
Current Discovery
Methods
Low Efficacy
Single Markers
Ciphergen
ProteinChip
Technology
High Efficacy
Multi-markers
0
0.5
Biomarker Discovery/ID
1
Ab Dev.
1.5
Assay Dev.
2
2.5
Validation
3
HT Screen
3.5