Biomarker development for
targeted cancer therapeutics,
a real life story
ODDP 2015, Amsterdam
Prof. Alain van Gool
Professor Personalized Healthcare
Coordinator Radboudumc Technology Centers
Head Radboud Center for Proteomics, Glycomics and Metabolomics
Senior Scientist Integrator Biomarkers
Based on data and slides from projects @Organon, Schering-Plough, MSD
2
8 years academia (NL, UK)
(molecular mechanisms of disease)
13 years pharma (EU, USA, Asia)
(biomarkers, Omics)
4 years applied research institute (NL, EU)
(biomarkers, personalized health)
4 years university medical center (NL)
(personalized healthcare, Omics, biomarkers)
My background
1991-1996
(PhD)
1996-1998
(post-doc)
2009-2012
(visiting prof)
1999-2007 2007-2009 2009-2011
2011-now
(prof)
2011-now
2
3
Agenda
Background
– Personalized medicine
– Need for biomarkers in oncology
Case study
– Biomarkers to support development of BRAF inhibitors for melanoma
4
Translational medicine in pharma
Basic Research
In Vitro Studies
Target Validation
Animal Models
Phase I and Phase II
-PoC- Studies
Phase III Studies
Clinical Research
Forward Translation Forward Translation
Reverse Translation Reverse Translation
(View drug development
as customer)
(Feed back clinical needs
and samples)
[van Gool et al, Drug Disc. Today 2010]
5
Biomarkers
Definition: ‘a characteristic that is objectively measured and evaluated as an
indicator of normal biological processes, pathogenic processes, or
pharmacologic responses to a therapeutic intervention’
Molecular biomarkers can provide a molecular impression of a biological
system (cell, animal, human)
Biomarkers can be various analytes:
PSA protein – blood, indicator of prostate cancer
Cholesterol – blood, risk indicator for coronary and vascular disease
{Biomarkers definition working group, 2001 }
MRI scan – shows abnormal tissue, like brain tumor
6
Biomarker strategy based on key questions
Does the compound get to the site of action?
Does the compound cause its intended pharmacological/
functional effects?
Does the compound have beneficial effects on disease or
clinical pathophysiology?
What is the therapeutic window (how safe is the drug)?
How do sources of variability in drug response in target
population affect efficacy and safety?
Lead
Optimization
Exploratory
Development PoC
Lead
Discovery
Target
Discovery
Exposure ?
Mechanism ?
Efficacy ?
Safety ?
Responders ?
Core of Biomarker Strategy and Development planning
Start in Early Discovery, expand in Lead Optimization, complete in clinical Proof of Concept
{Concept by de Visser and Cohen, CHDR}
{van Gool et al, Drug Disc. Today 2010}
7
Biomarker strategy: Data-driven decisions
To be made during testing of drug in preclinical and clinical disease models:
Target engagement? Effect on disease?
yes yes !
no no
• No need to test current
drug in large clinical trial
• Need to identify a more
potent drug
• Concept may still be
correct
• Concept was not correct
• Abandon approach
• Proof-of-Concept
• Proceed to full
clinical
development
“Stop early, stop cheap”
“More shots on goal”
Include personalized differences at every stage when possible.
8
Rational selection of best targets and drugs
works The 5R’s assessment:
• Right Target
• Right Tissue
• Right Safety
• Right Patients
• Right Commercial Potential
8
9
High attrition in oncology drug development
{Kola & Landis, Nat. Rev. Drug Disc. (2004) 8: 711}
10
Source: Arrowsmith: Nature Reviews Drug Discovery 2011
• Success rates of clinical proof-of-concept have dropped from 28% to 18% • Insufficient efficacy as the most frequent reason • Better therapies following Personalized Medicine strategies are needed • Key to apply translational biomarkers for personalized therapy
Need for Personalized Medicines
Analysis of 108 failures in phase II
Reason for failure Therapeutic area
10
11
Consider individual differences in life science research
12
{Source: Chakma. Journal of Young Investigators. 2009}
Principle of Personalized/Precision/Targeted Medicine
13 13 Alain van Gool, NanoNext.NL, 3 July 2015
Optimal Personalized / Precision / Targeted Medicine
14
Biomarker need in oncology clinical care
Early detection tumor
Determine mechanism of pathophysiology
Determine tumor stage
Early detection benign to malignant tumor progression
Detect residual disease after therapy
Early and sensitive detection metastatic circulating cells
Early detection metastatic tumor
Understand why people respond differently
Main needs:
Need for biomarkers to develop more targeted therapies
Need for biomarkers for patient selection
15
Biomarker need in oncology drug development
Determine mechanism of pathophysiology of tumor
Verify published data on drug target
Select and develop a drug with
– Sufficient selectivity
– Highest efficacy Lead Optimisation
– Lowest off-target safety risk
Test exposure, efficacy and safety of drug in preclinic model
Test exposure, efficacy and safety of drug in clinical trials
Test efficacy in stratified patients, selected on mechanism
Monitor drug efficacy and safety post-market introduction
Back-translation of clinical findings to research
Consistent application of translational biomarkers
16
Agenda
Background
– Personalized medicine
– Need for biomarkers in oncology
Case study
– Biomarkers to support development of BRAF inhibitors for melanoma
17
Case study: Development RAF inhibitors for melanoma
{Miller and Mihm,
2006}
18
Mechanism of pathophysiology in BRAF mutated tumors
V600E
Kinase domain
{Roberts and Der, 2007}
B-RAFV600E mutation: constitutively active kinase, oncogenic addiction
Overactivate ERK pathway drives cell proliferation
RAF inhibitors shown to block growth of tumors with B-RAFV600E mutation
Prevalence of B-RAFV600E
– Melanoma (60%), colon (15%), ovarian (30%), thyroid (30%) cancer
19
19
{Source: Prof Khusru Asadullah, Head of Global Biomarkers Bayer Healthcare}
20
Cellular efficacy by selective B-RAF inhibition by siRNA
Wild
type
Moc
k
BRAF
1
BRAF
5
CRAF
1
CRAF
3
ARAF
4
ARAF
8
siCont
rol
GFP
0
25
50
75
100
125
150
Percen
tag
e
Inhibition of RAF-MEK-ERK
pathway and induction of
apoptosis by siRNA (shown effect in A375 cells)
Inhibition of cell proliferation
by siRNA (shown effect in A375 cells)
G2/MS
G1
A0
A0 : 38 %
G1 : 42 %
S : 6 %
G2/M : 5 %
G2/M
G1
S
G1 : 65 %
S : 17 %
G2/M : 15 %
G1 : 56 %
S : 16 %
G2/M : 17 %
SG2
G1
G2S
G1
G1 : 65 %
S : 17 %
G2/M : 12 %
B-RAF C-RAF
A-RAF GFP
G2/MS
G1
A0
A0 : 38 %
G1 : 42 %
S : 6 %
G2/M : 5 %
G2/MS
G1
A0
A0 : 38 %
G1 : 42 %
S : 6 %
G2/M : 5 %
G2/M
G1
S
G1 : 65 %
S : 17 %
G2/M : 15 %
G2/M
G1
S
G1 : 65 %
S : 17 %
G2/M : 15 %
G1 : 56 %
S : 16 %
G2/M : 17 %
SG2
G1
G1 : 56 %
S : 16 %
G2/M : 17 %
SG2
G1
SG2
G1
G2S
G1
G1 : 65 %
S : 17 %
G2/M : 12 %
G2S
G1
G1 : 65 %
S : 17 %
G2/M : 12 %
B-RAF C-RAF
A-RAF GFP
Induction of apoptosis
by siRNA
(shown effect in A375 cells)
Key response selection biomarker is B-RAFV600D/E mutation
Key pathway biomarker is phosphorylated ERKSer202/204 = p-ERK
B - RAF
C - RAF
ERK
B - actin
A - RAF
Mock GFP Si -
control C - RAF A - RAF
PARP
B - RAF WT
p-MEK
p-ERK
-
Mock GFP Si -
control C - RAF A - RAF B - RAF WT
21
Cellular efficacy by RAF kinase inhibitor compounds
Inhibition of proliferation (A375, SK-MEL-24, Colo-205)
Inhibition of anchorage-
independent growth
in soft agar (A375)
Inhibition of RAF-MEK-ERK pathway (A375, SK-MEL-24, Colo-205)
Sorafenib
Sorafenib CI1040 Org240390 SB590885 Org245224 Org245108
Solvent No compound
Sorafenib CI1040 Org240390 SB590885 Org245224 Org245108
A375
Cells:
Compounds:
SK-MEL-24
Colo-205
Sorafenib
(multikinase)
CI-1040
SB 590885
Perc
en
tag
e g
row
th
- 10 - 9 - 8 - 7 - 6 - 5 - 4 - 10
0 10 20 30 40 50 60 70 80 90
100 110 120
Log conc. (M)
CI-1040
(MEK)
SB590885
(B-RAF)
22
Analysis ERK pathway activity
A375 treated with MEKi #1 A375 treated with RAFi #1
RSK RSK RSK
p-MEK
p-ERK
p-RSK
-10 -8 -6
0
50
100
150
DM
SO
Log [SCH 772984, M]
% o
f E
RK
ph
os
ph
ory
lati
on
-10 -8 -6 0
50
100
150
DM
SO
Log [SCH 772984, M]
% o
f M
EK
ph
os
ph
ory
lati
on
-10 -8 -6
0
50
100
150
DM
SO
Log [SCH 772984, M]
% o
f R
SK
ph
os
ph
ory
lati
on
Log [ , M]
Log [ MEKi #1 , M]
MEKi #1
IC50 = 35.70 nM
IC50 = 14.26 nM
No inhibition
Concentration MEKi #1 Concentration RAFi #1
Immunoassays to monitor phosphorylation biomarkers in ERK pathway
(ELISA, western blotting, mass spectrometry, reverse phase protein arrays)
23
Example flow chart B-RAF Lead Optimisation
IC50 < 20 nM IMAP
Dose-dependent inhibition
IC50 < 100nM in 2 out of 3
B-RAFV600E cell lines
IC50 < 100 nM P-ERK P-ERK cellular
Phase 1a
Phase 1b
Solubility, eLogD,
Pampa B-RAF biochemical
In vitro ADME
Cell proliferation
ADME-PK data in range to
allow 1 or 2x daily dosing
EDC selection
Phase 2a
Phase 2b
Selection phase
PK rodent
Selectivity 20 kinases
In vitro safety
Pilot xenografts Full kinase profiling
PK
dog/monkey
14 day rat tox,
pilot Ames, novascreen,
CV safety, phototox
Pilot CMC
Reduced tumor growth at
equivalent of anticipated
human dose
Efficacy assays
ADME-Tox assays
Decisive path
Safety profile supportive
of therapeutic window
Cell apoptosis
Xenograft mouse models ?
24
A375_P_ERKpEC50
5.5
6
6.5
7
7.5
5.5 6 6.5 7 7.5
Activity B-RAF inhibitors in melanoma cell line
Proliferation (pIC50)
Pathway inhibition (pIC50)
Lead
Competitor compound
Best own compound
Overlapping
two week S&T cycles:
Week 1: Synthesis
Week 2: Testing
Start with 1 lead, and
S&T of up to 1000
derivatives.
25
Structure Activity Relationships
NH
NH
O
O
O
NH
O
CN
N
S
R
N
NS
R1
R2
(lead)
Central phenyl: only m-F allowed
Allosteric backpocket:
• aryl required
• meta subst required
• heteroaromates allowed
• solubilizers allowed
Linker:
• NHCO most active
• CONH, urea are allowed
• alkylated amide not allowed
Linker:
• NHCO most active
• variation allowed but 10-50 fold loss
Lead compound modeled in crystal structure of B-RAF kinase domain
Hinge: • subst of benzoxizanone
optimal for cellular activity
• solubilizers allowed
• other scaffolds allowed:
• substituted thienopyrazines
most optimal
26
Kinase selectivity
Medium screening of >200 kinases using biochemical assays
Read-out = phosphorylation of substrates
Limited translational value but selection of potential off-target hits
Subsequent validation needed on cellular level
% inihibition
-60
-40
-20
0
20
40
60
80
100
120
Example:
Kinase selectivity of 3 compounds tested
under the same assay conditions
% inhibition is shown; each dot is one assay
27
Cancer cell line panel testing ORG RAFi Proliferation EC50 vs. RAF Genotype
1
10
100
1000
10000
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
Pro
lif E
C5
0 (
nM
)
Braf V600D Hetero
Braf V600E Homo
Braf V600E Hetero
Braf WT
CHIR 265 Prolif EC50 vs. RAF Genotype
1
10
100
1000
10000
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
Pro
lif E
C5
0 (
nM
)
Braf V600D Hetero
Braf V600E Homo
Braf V600E Hetero
Braf WT
AZD6244 Prolif EC50 vs. RAF Genotype
1
10
100
1000
10000
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
Pro
lif E
C5
0 (
nM
)
Braf V600D Hetero
Braf V600E Homo
Braf V600E Hetero
Braf WT
772984 Prolif EC50 vs. RAF Genotype
1
10
100
1000
10000
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
Pro
lif
EC
50
(n
M)
Braf V600D Hetero
Braf V600E Homo
Braf V600E Hetero
Braf WT
RAFi 1 MEKi
RAFi 2 ERKi
Efficacy compound 1-4 linked to BRAFV600D/E mutational status
Compound 1 limited effect in WT cell lines (less off target effects?)
28
Discovery of improved biomarkers for RAF inhibitors
Aim: identify soluble protein biomarker in blood that reflects
inhibition of ERK pathway in tumor with B-RAFV600D/E mutation
(More practical than p-ERK protein analysis in tumor biopsy)
(Enabling personalized medicine)
Pharmacogenomics approach:
– A375 melanoma cells
– Homozygote BRAFV600E mutation
– Robust model system for method development
– Investigate effect of 7 inhibitors
• 4x RAFi
• 2x MEKi
• 1x ERKi
on gene expression, proliferation, apoptosis, etc
29
Pharmacogenomics in A375 melanoma cells
• Efficient approach
• Highly reproducible data with
robust gene modulation
• Identify compound-specific and
common differential transcripts
• Select candidate biomarkers
RAFi #4
MEKi #1MEKi #2
RAFi #3
RAFi #1
RAFi #2
ERKi #1
RAFi #4
MEKi #1MEKi #2
RAFi #3
RAFi #1
RAFi #2
ERKi #1
RAFi #4
MEKi #1MEKi #2
RAFi #3
RAFi #1
RAFi #2
ERKi #1
RA
Fi
#1
RA
Fi
#2
RA
Fi
#3
RA
Fi
#4
ME
Ki
#1
ME
Ki
#2
ER
Ki
#1
RA
Fi
#1
RA
Fi
#2
RA
Fi
#3
RA
Fi
#4
ME
Ki
#1
ME
Ki
#2
ER
Ki
#1
RAFi #1
RAFi #2
RAFi #4
RAFi #1
RAFi #2
RAFi #4
Data for RAFi #4
4x RAFi
2x MEKi
1x ERKi
30
• ~200 genes with >10 fold change.
• Overlap and differences between compound-regulated genes
• Methods applied to select new candidate biomarkers for validation, e.g. as
secreted proteins in plasma
• Selection of ERK pathway responsive transcripts, e.g. IL-8
Selection biomarkers from pharmacogenomics A375 cells
RA
Fi
#4
RA
Fi
#1
RA
Fi
#2
ER
Ki
#1
RA
Fi
#3
ME
Ki #
1
ME
Ki #
2
DM
SO
31
Zoya R. Yurkovetsky, John M. Kirkwood et al. Clin Cancer Res 2007;13(8) April 15, 2007
123 pg/ml
9 pg/ml
p < 0.001
Determination of IL-8 levels (one of 29 serum cytokines analyzed) in
179 melanoma patients (stage II & III) & 379 healthy individuals
Elevated levels of IL-8 in Patients with Melanoma
32
Validation study to confirm IL-8 in melanoma
Tissue Plasma
Normal Healthy Controls 40 50
Stage 1 11 11
Stage 2 11 11
Stage 3, non-metastatic 4 4
Stage 3, metastatic 11 11
Stage 4, non-metastatic 3 3
Stage 4, metastatic 19 19
Stage 1 Stage 2 Stage 3 Stage 4
H&E staining; 20x
Clinical samples used (from two independent commercial biobanks)
33
Validation study to confirm IL-8 in melanoma
Stage 1 Stage 2 Stage 3 Stage 4
H&E staining; 20x
Analysis done:
• Genetic analysis for BRAFV600E/D mutation in genomic DNA from tissue samples
• IL-8 mRNA analysis in tissue samples by in situ hybridisation using bDNA probes
(multiplexing with 12 ERK pathway response transcripts)
• IL-8 protein analysis in tissue samples by immunohistochemistry (in parallel with 4 other
ERK pathway response proteins, Ki67, Tunnel)
• IL-8 protein analysis in matching plasma and serum by IL-8 immunoassay (3 formats:
ELISA, Luminex, Mesoscale; singleplex and multiplex)
• Statistical data analysis
34
Plasma IL-8 levels vs Melanoma Stages
No confirmation of literature: no change in IL-8 protein levels in plasma
samples of melanoma patients. Reason?
35
No change in plasma & serum IL-8 levels in melanoma
Serum IL-8 levels in various Stages of Melanoma
Healthy control (n=10) Melanoma (n=37)
0
20
40
60
80
Me
an
IL
-8 l
ev
els
(p
g/m
l)
Plasma IL-8 levels in various Stages of Melanoma
Healthy control (n=20) Melanoma (n=59)
0
5
10
15
20
Me
an
IL
-8 l
ev
els
(p
g/m
l)
No confirmation of literature: no change in IL-8 protein levels in melanoma
Reason?
Conclusion:
Key response selection biomarker is B-RAFV600D/E mutation
Key pathway biomarker is phosphorylated ERKSer202/204 = p-ERK
36
Alignment with: - Experimental medicine
- Competitive intelligence
- Strategy
- Toxicology
- Formulation
- External experts (clinics, academics)
Predict clinical efficacy in oncology
Cells
Cell line xenografts (PoM, PoP)
Healthy subjects (PoM)
Cancer patients (PoM, PoP)
Selected cancer patients (PoC)
PoM – Proof of Mechanism
PoP – Proof of Principle
PoC – Proof of Concept
Primary tumor xenograft models
Genetically engineered mouse models
(PoM, PoP, non-pivotal PoC)
37
Clinical efficacy of Vemurafenib, a novel BRAF inhibitor
Key biomarkers:
Exposure: -
Mechanism: p-ERK, Cyclin-D1
Efficacy: Ki-67, 18FDG-PET, CT
Safety: -
Selection: BRAFV600E mutation
Clinical endpoint: progression-free survival (%)
{Source: Flaherty et al, NEJM 2010} {Source: Chapman et al, NEJM 2011}
38
History of Zelboraf (Vemurafenib)
Davis M J , Schlessinger J J Cell Biol 2012;199:15-19
© 2012 Davis and Schlessinger
39
Clinical effects of Vemurafenib
{Wagle et al, 2011, J Clin Oncol 29:3085}
Before Rx Vemurafenib, 15 weeks Vemurafenib, 23 weeks
• Strong initial effects vemurafenib
• Drug resistancy
• Reccurence of tumors
40
People are complex systems …
41 {Source: Yancovitz, PLoS One 2012}
Tumor tissue heterogeneity
• BRAFV600D/E is the driving
mutation in melanoma
• However, also no BRAFV600D/E
mutation found in parts of a
primary melanoma
• Molecular heterogeneity in
diseased tissue
• Biomarker levels in tissue will
vary
• Biomarker levels in body
fluids will vary
• Major challenge for
(companion) diagnostics
42
Biomarker innovation gaps
Discovery Clinical
validation/confirmation
Diagnostic
test
Number of
biomarkers
Gap 1
Gap 2
Gap 3
43
Biomarker innovation gaps: some numbers
Discovery Clinical
validation/confirmation
Diagnostic
test
Number of
biomarkers
Gap 1
Gap 2
Gap 3
5 biomarkers/ working day
1 biomarker/ 1-3 years
1 biomarker/ 3-10 years
?
Eg Biomarkers in time: Prostate cancer May 2011: n= 2,231 biomarkers Nov 2012: n= 6,562 biomarkers Oct 2013: n= 8,358 biomarkers Nov 2014: n= 10,350 biomarkers Oct 2015: n = 11,856 biomarkers
44
Reasons for biomarker innovation gap
• Not one integrated pipeline of biomarker R&D
• Publication pressure towards high impact papers
• Lack of interest and funding for confirmatory biomarker
studies
• Hard to organize multi-lab studies
• Biology is complex on organism level
• Data cannot be reproduced
• Bias towards extreme results
• Biomarker variability
• …
{Source: John Ioannidis, JAMA 2011}
{Source: Khusru Asadullah, Nat Rev Drug Disc 2011}
45
Build biomarker validation pipelines
Standardisation, harmonisation, knowledge sharing needed in:
1. Assay development
2. Clinical validation
3. Regulatory acceptance
46
Agenda
Background
– Personalized medicine
– Need for biomarkers in oncology
Case study
– Biomarkers to support development of BRAF inhibitors for melanoma
Take home messages:
Choose and validate your biomarkers wisely
Collaborate
Realize human biology is complex
47
Thanks to:
Biomarker strategies Collaborators
Members of:
- Organon Biomarker Platform
- Schering-Plough Biomarker Group
- Merck Research Labs - Molecular Biomarkers
Translational Medicine Research Centre Singapore
Colleagues, particularly:
Erik Sprengers, Shian-Jiun Shih, Brian Henry, Hannes
Hentze, Zaiqi Wang, Rachel Ball, Meena Krishnamoorthi,
Aveline Neo, Sabry Hamza, Nicole Boo, Lee Kian-Chung,
Vidya Anandalaksmi
MSD/Merck
Colleagues, particularly in:
- Oss (Netherlands)
- Rahway, Kenilworth, Boston (East Coast, USA)
- San Francisco, Palo Alto (West Coast, USA)
Many in Asia, Europe, USA, including:
- Academic
- Consortia
- Contract research organizations
- Vendors
Saco de Visser, Adam Cohen Centre for Human Drug Research, Leiden, NL