Transcript ASCO_2009_files/Benson BiomarkerDrivenTrials ASCO 2009

Biomarker-Driven Design:
Complexities Using A
Colon Cancer Model
Al B. Benson III, MD, FACP
Professor of Medicine
Associate Director for Clinical Investigations
Robert H. Lurie Comprehensive Cancer Cente
of Northwestern University
Prognostic Markers versus
Predictive Markers
• Markers may have both prognostic and
predictive value
– This can complicate assessment
Prognostic marker
Indicates the likelihood of outcome (tumor recurrence or patient
survival) regardless of the specific treatment the patient receives
Predictive marker
Indicates the likelihood of response to a specific therapy
www.cancerdiagnosis.nci.nih.gov
Prognostic versus Predictive Markers
Comparative Effectiveness
• Individual factors contribute to differences
in clinical outcomes
– Race or ethnic diversity
– Co-morbidities
– Drug-drug interactions
– Tumor heterogeneity
– Tumor genetics
– Host genetics
All patients with same diagnosis
Alternate therapy
non-responders
and toxic responders
Standard therapy
Responders and Patients
Not Predisposed to Toxicity
Why Correlational Studies in
Colorectal Cancer?
• Trials represent a “generic” population
– Predictably a high % will have no benefit
• Tumors are heterogenous
• Numerous new “targeted” therapies, e.g.,
EGFR, VEGF
• Models: Breast cancer, GIST
• Toxicities
From “Marker” to “Test”
• Significant and independent value
• Validated by clinical testing
• Feasibility, reproducibility and widely
available with quality control (robust)
• Performance should benefit the patient
Ann Thor, ECOG, 2002
Comparison and Applicability of Different
Methodologies for Assessment of Tumor Markers
PCR
PCR
SSCP
LOH
DNA
DNA
DNA
RTPCR
mRNA
Use in formalinfixed, waxembedded tissue
Y
Y
Y
N
N
Y
Y
Microdissection
needed
Y
Y
Y
Y
Y
N
N
Cellular
localization
evaluable
N
N
N
N
N
Y
Y
Application to
routine diagnosis
Y
N
N
N
N
Y
Y
Cellular
constituent
examined
Northern
blotting
mRNA
ISH
ICH
DNA / Protei
mRNA
n
McLeod HL and Murray GI. British J of Cancer 79(2)191-203, 1999
Prevalence of Alterations
Prevalence (%95%CI)
100
80
60
40
20
0
18qL0H
17pL0H
p53
overexpr.
p21waf1
expr.
8pL0H
Current
A
All patients receive
standard treatment (A)
patients
B
Clinical trials
survival benefit from A
Future
A
B
patients
Molecular analysis
of tumor and patients
C
D
Choice of treatment
dependent upon molecular
profile of tumor and on
patient genotype
Patient biology
Lymph node status
Tumor biology
Cancer
Outcome
Distant metastasis
Surgical technique
Access to care
All patients with same diagnosis
All patients with same diagnosis
Alternate therapy
non-responders
and toxic responders
Standard therapy
Responders and Patients
Not Predisposed to Toxicity
Marker Analyses from Clinical Trials
• Retrospective Analyses
– Majority of marker reports
– Incomplete tissue collection
– Small numbers of patients
– Various methodologies
– Can be hypothesis generating
– Exception = Kras
Marker Analyses from Clinical Trials
• Prospective Correlative Studies in Clinical
Trials
– Tissue collection not mandated
– Statistically significant number of patients and
comparisons
– Robust clinical data
– Many trials now include correlatives
Marker Analyses from Clinical Trials
• Marker-driven Treatment Strategy
– Stratification
– Treatment assignment
Incidence of Colorectal Cancer
U.S. 2003 N=152,000
Stage I
24%
Stage IV
22%
Stage II
26%
Stage III
29%
Eligible for
Adjuvant
Chemotherapy
N=83,000 (55%)
th
6
AJCC
Edition:
Colorectal Cancer
- Stage II divided
into
• IIA (T3N0M0)
• IIB (T4N0M0)
- Stage III divided into
• IIIA (T1-2N1M0)
• IIIB (T3-4N1M0)
• IIIC (TanyN2M0)
Estimates of 5 Year DFS (%) with
Surgery Plus Adjuvant Therapy
Nodal
Status
T stage
Low Grade
High Grade
S
+AT
S
+AT
T3
T4
73
60
77
66
65
51
70
57
1-4 nodes
T1-T2
T3
T4
62
49
33
75
65
51
53
38
23
68
56
40
> 5 nodes
T1-T2
T3
T4
39
24
11
57
43
27
28
15
5
46
32
17
0 nodes
Adapted from Cill et al.. J Clin Oncol 22 :1801, 2004
MOSAIC: Treatment arms
FOLFOX4: LV5FU2 + Oxaliplatin 85mg/m²
D1 5FU bolus
LV
5-FU infusion*
D2 5FU bolus
LV
LV
5-FU infusion*
OXA
R
D1 5FU bolus
LV5FU2
LV
5-FU infusion*
D2 5FU bolus
LV
5-FU infusion*
Every 2 weeks, 6 months of treatment (12 cycles)
Disease-free Survival: ITT
1.0
0.9
p=0.003
0.8
5.9%
Probability
0.7
0.6
0.5
Events
0.4
0.3
0.2
FOLFOX4
304/1123 (27.1%)
LV5FU2
360/1123 (32.1%)
FOLFOX4
LV5FU2
HR [95% CI]: 0.80 [0.68–0.93]
0.1
0
0
6
Data cut-off: June 2006
12
18
24
30
36
42
48
Disease-free survival (months)
54
60
Disease-free Survival: Stage II
and Stage III Patients
1.0
p=0.258
0.9
3.8%
0.8
p=0.005
Probability
0.7
7.5%
0.6
0.5
0.4
0.3
0.2
0.1
HR [95% CI]
p-value
Stage II
0.84 [0.62–1.14]
0.258
Stage III
0.78 [0.65–0.93]
0.005
FOLFOX4 stage II
LV5FU2 stage II
FOLFOX4 stage III
LV5FU2 stage III
0
0
6
Data cut-off: June 2006
12
18
24
30
36
Months
42
48
54
60
66
72
Disease-free Survival: High-risk
Stage II Patients
1.0
0.9
0.8
Probability
0.7
7.2%
FOLFOX4 n=286
0.6
LV5FU2
n=290
0.5
0.4
3-year
5-year
0.3
FOLFOX4
85.4%
82.1%
0.2
LV5FU2
80.4%
74.9%
HR [95% CI]: 0.74 [0.52–1.06]
0.1
High-risk stage II- defined as at least
one of the following:
T4, tumor perforation, bowel obstruction,
poorly differentiated tumor, venous
invasion , <10 lymph nodes examined;
Data cut-off: June 2006
0
0
6
Exploratory analysis
12
18
24
30
36
42
48
54
Disease-free survival (months)
60
66
72
Approximate Number of Patients Needed
to Detect a Realistic
Treatment Benefit*
Dukes’ B
Survival
No. of
ARR
Dukes’ C
Patients
Survival
65%
At 3 years
85%
2.5%
8,000
At 4 years
80%
3.3%
5,800
At 5 years
75%
4.0%
4,700
No. of
ARR
Patients
5.2%
3,400
58%
6.0%
2,800
50%
6.6%
2,400
Abbreviation: ARR = absolute risk reduction
•For 90% power of detecting the treatment benefit using two-tailed significance tests at the 5% level,
assuming the true relative risk reduction is 18% for both Dukes’ B and Dukes’ C.
Buyse, Piedbois, 2001
Prognostic Factors in Colorectal Cancer
COLLEGE OF AMERICAN PATHOLOGISTS CONSENSUS
Category 1 – evidence from multiple statistically-robust
published trials and used in pt. management
Category IIA – extensively studied and sufficient for path
reports, but needs validation
Category IIB – promising
Category III – insufficient study
Category IV – well-studied and no prognostic
significance
Prognostic Factors in Colorectal Cancer
COLLEGE OF AMERICAN PATHOLOGISTS CONSENSUS
Category I

path-local extent of tumor = pT

path-nodes = pN

blood or lymphatic invasion

post-op residual tumor = R (e.g., + margin)

post-op  CEA
Category IIA

tumor grade

radial margin status

residual tumor s/p neoadjuvant tx
Intergroup Adjuvant Colon Cancer
INT 0035 (E 2284)
S
U
Observation
R
G
Levamisole
E
R
Y
5-FU /
levamisole
Intergroup Adjuvant Colon Cancer
INT 0089 (E 2288)
S
5-FU / leucovorin (Mayo)
U
R
5-FU / leucovorin (Roswell)
G
E
5-FU / levamisole
R
Y
5-FU / levamisole /
leucovorin
Analysis of Molecular Markers in Patients
with Stage III Colon Cancer
Watanbe T, et al. N Engl J Med 344(16);1196-1206, 2001
Analysis of Molecular Markers in Patients
with Stage III Colon Cancer
Watanbe T, et al. N Engl J Med 344(16);1196-1206, 2001
E5202 Trial Schema
High-Risk Patients
MSS/18q LOH or
MSI-L/18q LOH
are
Stratify:
RANDOMIZED
Disease stage
(IIA or IIB)
Microsatellite stability
(stable vs MSI)
18q LOH
Low-Risk Patients
MSS or MSI-L with
retention of 18q alleles
MSI-H
MSI-L = low-level microsatellite instability
MSI-H = high-level microsatellite instability
*Bevacizumab continued for an additional 6 months
Arm A:
mFOLFOX6
q2w × 12
Arm B:
mFOLFOX6 +
bevacizumab*
q2w × 12
Arm C:
Observation only
E5202 Trial Design: Sample
Submission
• Tumor and normal tissue sample required for
enrollment
– Samples must be formalin-fixed paraffin blocks or
unstained histologic sections
– Submission time points are crucial
• Received no later than 50 days following surgery
• Received within 5 days of trial registration
– Surgeons at participating institutions should be aware of
timeline in order to introduce patients
to trial
• Critical given timeline of tissue collection
E5202 Correlative Studies
• Correlate tumor biologic characteristics with
survival of patients treated with test regimens
– Microsatellite stability
– 18q LOH
• All tissue from study to be archived by ECOG
coordinating center and assessed for biologic
characteristics by MD Anderson laboratories
• Tissue from studies will be archived for future
assessment
Deficient Mismatch Repair as a
Predictive Marker for Lack of Benefit
from 5-FU based Chemotherapy in
Adjuvant Colon Cancer
DJ Sargent, S Marsoni, SN Thibodeau, R
Labianca, SR Hamilton, V Torri, G Monges, C
Ribic, A Grothey, S Gallinger
ASCO 2008
A quantitative multi-gene RT-PCR assay for
prediction of recurrence in stage II colon cancer:
Selection of the genes in 4 large studies and results
of the independent, prospectively-designed
QUASAR validation study
David Kerr1, Richard Gray2, Philip Quirke3, Drew Watson4,
Greg Yothers5, Ian Lavery6, Mark Lee4, Michael O'Connell5,
Steven Shak4, Norman Wolmark5 and the Genomic Health
& QUASAR Colon Teams
1. University of Oxford, Oxford, UK; 2. Birmingham Clinical Trials Unit,
Birmingham, UK; 3. Leeds Institute of Molecular Medicine, Leeds, UK; 4.
Genomic Health, Inc., Redwood City, CA; 5. National Surgical Adjuvant Breast
and Bowel Project, Pittsburgh, PA; 6. Cleveland Clinic Foundation, Cleveland, OH
The Need for Individualized Therapy in Stage II
Colon Cancer
• The challenge: Which stage II colon cancer patients should
be treated with adjuvant chemotherapy?
– 75-80% cured with surgery alone, but no method to identify them
– Absolute benefit of chemotherapy is small and no consensus in
guidelines on who to treat
– Chemotherapy has significant toxicity
• Today, decision to give chemotherapy subjectively based on:
– Clinical/pathologic markers of risk which are inadequate
• Not informative for majority of patients
– Patient age, co-morbidities, preferences
Development and Validation of a Multi-Gene RTPCR Colon Cancer Assay
Colon Cancer Technical Feasibility
• NSABP and CCF
Collaborations 761 genes studied in
1,851 patients to select
genes which predict
recurrence and/or
differential 5FU/LV
benefit
• Clinical Validation of
final assay in a large,
prospectively-designed
independent study
Development Studies
Surgery Alone
Development Studies
Surgery + 5FU/LV
NSABP C-01/C-02 (n=270)
NSABP C-04 (n=308)
CCF (n = 765)
NSABP C-06 (n=508)
Selection of Final Gene List & Algorithm
Validation of Analytical Methods
Clinical Validation Study – Stage II Colon Cancer
QUASAR (n=1,436)
Test Prognosis and Treatment Benefit
QUASAR RESULTS: Colon Cancer Recurrence
Score Predicts Recurrence Following Surgery
Prospectively-Defined Primary Analysis in Stage II Colon Cancer (n=711)
RECURRENCE SCORE
STROMAL
FAP
INHBA
BGN
CELL CYCLE
Ki-67
c-MYC
MYBL2
GADD45B
REFERENCE
ATP5E
GPX1
PGK1
UBB
VDAC2
Risk of recurrence at 3 years
Calculated from Tumor
Gene Expression
35%
Group Risk (by Kaplan-Meier)
22%
12%
18%
30%
25%
20%
15%
10%
p=0.004
5%
0%
0
10
20
30
40
50
Recurrence Score
60
70
QUASAR RESULTS: Recurrence Score, T Stage,
and MMR Deficiency are Key Independent Predictors
of Recurrence in Stage II Colon Cancer
Key
Variable
P
Category
HR value
Mismatch Repair (MMR) by IHC Deficient (13% of pts) 0.32 <.001
T4 (15% of pts)
1.83 0.005
Tumor Grade
High (29% of pts)
0.62 0.026
# Nodes Examined
<12 (62% of pts)
1.47 0.040
T Stage
Lymphovascular Invasion
RS per 25 units
Present (13% of pts) 1.40 0.175
Continuous
1.61 0.008
Multivariate Analysis
Summary and Conclusions
• The prospectively-defined continuous Recurrence Score has been
validated as a predictor of recurrence in stage II colon cancer patients
following surgery, and provides independent value beyond standard
measures of risk
• A separate score, based on a distinct set of 6 genes, was not
validated for prediction of differential 5FU/LV benefit
Implications for Clinical Practice
• The continuous RS provides individualized assessment of recurrence
risk and will have the greatest clinical utility when used in conjunction
with T stage and Mismatch Repair (MMR/MSI), particularly for the
majority of patients for whom those markers are uninformative (~70%
of pts)
• This is the first demonstration that a prospectively defined gene
expression assay can independently predict recurrence in colon
cancer
EGF-induced Signal Transduction
and Tumorigenesis
EGF
• Epidermal growth factor
EGFR
receptor (EGFR)
– A large tyrosine kinase
growth factor receptor
• Natural ligands
– TGF-, EGF
• Potential to block multiple
steps in the signal
transduction process
– Extracellular surface
– Intracellular targets
(+)
X
Anti-EGFR
RAS
pY
K K
RAF
SOS
pY
PI3-K
GRB2
MEK
pY
STAT
PTEN
AKT
MAPK
Gene transcription
Cell-cycle progression
G2
M
p27
S
X
X
Proliferation
Perez-Soler R. Oncologist. 2004;9:58-67.
G1
X
Survival/
anti-apoptosis
X
Angiogenesis
Invasion/
metastasis
Potential Biomarkers:
Methods of Testing
• EGFR protein expression
• EGFR gene copy number
• K-ras gene mutations
• EGFR ligands and phosphorylation
Fig 1. CONSORT diagram
Amado, R. G. et al. J Clin Oncol; 26:1626-1634 2008
Copyright © American Society of Clinical Oncology
Fig 2. Progression-free survival by treatment within KRAS groups
Amado, R. G. et al. J Clin Oncol; 26:1626-1634 2008
Copyright © American Society of Clinical Oncology
Fig 3. Subset analyses of progression-free survival in the KRAS wild-type group
Amado, R. G. et al. J Clin Oncol; 26:1626-1634 2008
Copyright © American Society of Clinical Oncology
Fig 4. Waterfall plots showing maximum percent decrease in target lesions (blinded central radiology)
Amado, R. G. et al. J Clin Oncol; 26:1626-1634 2008
Copyright © American Society of Clinical Oncology
Fig 5. Kaplan-Meier curves for overall survival by treatment and KRAS status
Amado, R. G. et al. J Clin Oncol; 26:1626-1634 2008
Copyright © American Society of Clinical Oncology
Fig A1. (A) Progression-free survival and (B) overall survival by KRAS status among patients receiving
panitumumab after progression on best supportive care alone
Amado, R. G. et al. J Clin Oncol; 26:1626-1634 2008
Copyright © American Society of Clinical Oncology