Genomic Health
Post-ASCO Clinical Data Review
June 2011
Steve Shak, MD
2
Outline
•
The Biology Century
– Principles for Evaluating Genomic Assays
•
Breast Cancer
– Tumor Biology Matters: Supporting Data for the Recurrence Score ® in Predicting
Neoadjuvant Therapy Response
– Can You Consistently Predict the Recurrence Score with One or a Combination of
Traditional Measures?
– Practice and Health Economic Analyses
•
Colon Cancer
– Confirmation Study (CALGB 9581)
•
Prostate Cancer
– Identification of Prostate Cancer Genes Associated with Clinical Recurrence
•
Final Thoughts
3
Cancer – The Biology Century
• Understanding and treating the underlying
biology
– Continual advances in technology
– Basic research on cancer stem cells (CSC)
– Circulating tumor cells
– Integration of genomic analysis in the clinical
context and the continuum of care
4
Evaluating Biomarkers for Clinical Use:
Key Principles
• Does the test deliver what patients, physicians, regulators, and
payors need?
– Most importantly, tests must be “Fit for Purpose” with evidence
relevant to that specific purpose
– Consistent results across multiple well-designed studies are required to
provide evidence for analytic performance, clinical validity, and clinical
utility (see Roadmap to Establish Clinical Utility)
– Test must be shown to have value beyond traditional measures
• Has the test been brought to a standardized implementation?
And has the evidence which supports its use been obtained in that
standardized implementation?
Hayes DF. Am Soc Clin Oncol Ed Book. 30-34, 2008; Simon R. J Clin Oncol. 2005;23:7332-7341.
5
Roadmap to Establish Clinical Utility
Define the
purpose
Technical
feasibility
Development
studies
Clinical validation
studies, including
comparative
effectiveness
Analytical
methods
validation
Analytical
methods
finalization
Treatment
decision
studies
Health
economic
analysis
6
Key Questions When Evaluating
Genomic Classifiers
Strongly
prognostic?
Accurate
and
reliable?
Predictive of
chemotherapy
benefit?
Fit for
purpose
Incorporated
in treatment
guidelines?
What is the
level of
evidence?
7
Selected Oncotype DX®
Breast Abstracts at ASCO
•
Tumor Biology Matters: Supportive Data for the Recurrence Score® in
Predicting Neoadjuvant Therapy Response
– Masuda – RS as Predictor of Clinical Response to Neoadjuvant Exemestane in
ER+ BC
– Peacock – Exploratory Oncotype DX Assessment in HER2-Negative BC Receiving
Ixabepilone-Cyclophosphamide Neoadjuvant Therapy
•
You Cannot Consistently Predict the Recurrence Score
– Liebermann – Evaluation of RS and Traditional Clinical-Pathologic Assessments in a
Large ER+, Node-Negative Cohort (Israel)
•
Practice & Health Economic Analyses
– Chen – Evaluating Utilization Characteristics for Oncotype DX in ESBC (USO
Network)
– Hassan – Cost Benefit Analysis of 21-gene assay in Canadian healthcare system
RS, Recurrence Score
Tumor Biology Matters:
Supportive Data for the
Recurrence Score® Result in
Predicting Neoadjuvant Therapy
Response
A Study of the Recurrence Score® by the
21-Gene Signature Assay as a Predictor of
Clinical Response to Neoadjuvant
Exemestane for 24 Weeks in Estrogen
Receptor-Positive Breast Cancer
N. Masuda, M. Toi, T. Ueno, T. Yamanaka, S. Saji, K. Kuroi, N. Sato, H. Takei, Y. Yamamoto,
S. Ohno, H. Yamashita, K. Hisamatsu, K. Aogi, H. Iwata, S. Saji, H. Sasano
National Hospital Organization Osaka National Hospital, Osaka, Japan; Breast Surgery, Kyoto University, Kyoto,
Japan; Kyoto University Hospital, Kyoto, Japan; National Kyushu Cancer Center, Fukuoka, Japan; Saitama Medical
University, International Medical Center, Saitama, Japan; Tokyo Metropolitan Cancer and Infectious Diseases Center,
Komagome Hospital, Tokyo, Japan; Niigata Cancer Center Hospital, Niigata, Japan; Saitama Cancer Center,
Saitama, Japan; Department of Breast and Endocrine Surgery, Kumamoto University, Kumamoto, Japan; Nagoya
City University Graduate School of Medical Sciences, Nagoya, Japan; Hiroshima City Asa Hospital, Hiroshima,
Japan; National Hospital Organization Shikoku Cancer Center, Eihme, Japan; Aichi Cancer Center Hospital, Nagoya,
Japan; Japanese Foundation for Multidisciplinary Treatment of Cancer, Tokyo, Japan; Tohoku University Hospital,
Sendai, Japan
Masuda N, et al. ASCO 2011. Abstract 558 (poster presentation).
10
Objectives
•
This exploratory study was conducted to investigate the clinical usefulness of
the 21-gene signature assay (Oncotype DX®) for prediction of the response
to neoadjuvant Exemestane therapy in postmenopausal patients with
hormone receptor-positive locally advanced stage II or IIIA breast cancer
•
Primary objective
– Association between the Recurrence Score® (RS) determined on core biopsy
specimens obtained at baseline and clinical response (defined as ≥50% reduction
in area on imaging after 24 weeks)
•
Secondary objectives
– Association of baseline quantitative ER by RT-PCR and Ki-67 by Central IHC with
clinical response
– Association of changes from baseline to post-treatment values of RS, quantitative
ER by RT-PCR and Ki-67 by Central IHC with clinical response
– Association of baseline RS, quantitative ER by RT-PCR and Ki-67 by Central IHC
with breast-conserving surgery (BCS) vs. mastectomy
Masuda N, et al. ASCO 2011. Abstract 558 (poster presentation).
11
Primary Analysis: Core Biopsy Recurrence
Score® Result and Clinical Response
RS Group
Clinical Response
Low (< 18)
19/32 = 59.4%
Intermediate (18 - 30)
10/17 = 58.8%
High (≥ 31)
3/15 = 20.0%
p=0.015 for Comparison Between Low and High RS Groups
(2-sided Fisher’s exact test)
RS, Recurrence Score
Masuda N, et al. ASCO 2011. Abstract 558 (poster presentation).
12
Logistic Regression Analyses of Baseline Core
Biopsy RS in Predicting Clinical Response
Model
Odds Ratio (95% CI)
p-value
Int. vs. Low
0.977 (0.296, 3.233)
0.970
High vs. Low
0.171 (0.040, 0.728)
0.017
2
Continuous RS/50
0.205 (0.044, 0.946)
0.042
3
Age (years)
1.012 (0.926, 1.106)
0.795
Tumor Size (mm)
1.033 (0.960, 1.112)
0.387
Continuous RS/50
0.172 (0.034, 0.858)
0.032
1
•
•
Effect
RS Group
When analyzed by RS groups, the low and intermediate RS groups had similar odds of clinical
response, whereas the high RS group had significantly lower odds of clinical response
When analyzed as a continuous score, a lower RS was significantly associated with a higher odds
of clinical response in both unadjusted and covariate-adjusted (for age and tumor size) analyses.
RS, Recurrence Score®
Masuda N, et al. ASCO 2011. Abstract 558 (poster presentation).
13
Logistic Regression Analysis of Baseline Core
Biopsy ER by RT-PCR and Ki-67 by IHC in
Predicting Clinical Response
Model Effect
•
•
Odds Ratio (95% CI)
p-value
4a
Continuous ER by RT-PCR
1.436 (0.963, 2.141)
0.076
4b
Age (years)
0.980 (0.897, 1.072)
0.665
Tumor Size (mm)
1.017 (0.950, 1.090)
0.624
Continuous ER by RT-PCR
1.480 (0.975, 2.246)
0.066
5a
Continuous Ki-67 by IHC
0.981 (0.948, 1.015)
0.273
5b
Age (years)
1.023 (0.934, 1.121)
0.627
Tumor Size (mm)
1.025 (0.955, 1.100)
0.502
Continuous Ki-67 by IHC
0.976 (0.940, 1.013)
0.197
The association between quantitative ER by RT-PCR and clinical response was of borderline statistical significance
Baseline Ki-67 by IHC was not associated with clinical response
Masuda N, et al. ASCO 2011. Abstract 558 (poster presentation).
14
Logistic Regression Analyses of Changes in
Markers From Core Biopsy to Resection in
Predicting Clinical Response
Model
Variable
Odds Ratio (95% CI)
p-value
6
Change from Baseline to
Post-Treatment RS/50
0.198 (0.013, 2.952)
0.240
7
Change from Baseline to
Post-Treatment ER by RT-PCR (CT)
0.753 (0.419, 1.354)
0.343
8
Change from Baseline to
Post-Treatment Ki-67 by IHC (%)
1.011 (0.967, 1.057)
0.629
Changes from pre-treatment to post-treatment values in Recurrence Score®, ER by RTPCR or Ki-67 by central IHC were not associated with clinical response
RS, Recurrence Score
Masuda N, et al. ASCO 2011. Abstract 558 (poster presentation).
15
Logistic Regression of Baseline Core Biopsy
Markers in Predicting Breast Conserving
Surgery (BCS)
Unadjusted
Adjusted for Tumor Size
and PgR Allred Score
Marker
Odds Ratio (95% CI)
pvalue
Odds Ratio (95% CI)
pvalue
RS/50
0.055 (0.009, 0.323)
0.001
0.016 (<0.001, 0.259)
0.004
ER by
RT-PCR
1.786 (1.150, 2.774)
0.001
1.881 (1.090, 3.245)
0.023
Ki-67 by IHC
0.957 (0.921, 0.994)
0.024
0.953 (0.907, 1.002)
0.060
•
Lower Recurrence Score® result, higher ER by RT-PCR, and lower Ki-67 by central IHC
were all significantly associated with greater odds of breast-conserving surgery
•
When adjusted for the effects of tumor size and PgR Allred Score, RS and ER by
RT-PCR maintained their significance, and Ki-67 was of borderline significance
RS, Recurrence Score
Masuda N, et al. ASCO 2011. Abstract 558 (poster presentation).
16
Recurrence Score® Result, Ki-67
and Prediction of BCS
Relationship Between Core Biopsy RS
and Core Biopsy Ki-67 by IHC
Core Biopsy RS
and Breast Conserving Surgery
RS Group
BCS
Low (< 18)
29/32 = 90.6%
Intermediate (18 - 30)
13/17 = 76.5%
High (≥ 31)
7/15 = 46.7%
p=0.002 for Comparison Between Low and
High RS Groups (2-sided Fisher’s exact test)
Multivariable Analysis of Core Biopsy RS
and Ki-67 and Breast Conserving Surgery
Effect
Odds Ratio (95% CI)
p-value
RS/50
0.048 (0.004, 0.514)
0.012
Ki-67 by IHC
1.005 (0.949, 1.064)
0.868
RS, Recurrence Score
Masuda N, et al. ASCO 2011. Abstract 558 (poster presentation).
•
Recurrence Score and Ki-67 by central
IHC are moderately correlated
(Spearman correlation coefficient = 0.672,
95% CI 0.506-0.785)
•
When Recurrence Score and Ki-67 by
central IHC were included in the same
logistic regression model, Recurrence
Score was the only statistically significant
predictor of breast-conserving surgery.
17
Summary and Conclusions
•
Pre-treatment Recurrence Score® result from the core biopsy predicts
clinical response to 24 weeks of neoadjuvant exemestane
– Approximately 60% of the patients in the low and intermediate RS groups experienced
a clinical response as compared with 20% in the high RS group
– The significance of the associations persisted when adjusting for traditional baseline
factors
•
A low RS in the core biopsy is significantly associated with a greater
likelihood of undergoing BCS which carries an overall better outcome for
patients (lower morbidity and better cosmesis than mastectomy)
•
Pre-treatment Ki-67 by central IHC was significantly associated with BCS,
but not when adjusted for the RS
– Changes from core biopsy to resection in RS, ER by RT-PCR and Ki-67 by central
IHC were not associated with clinical response
•
Neoadjuvant endocrine treatment may be an effective option in patients with
low or intermediate RS disease
RS, Recurrence Score
Masuda N, et al. ASCO 2011. Abstract 558 (poster presentation).
Ixabepilone and Cyclophosphamide as
Neoadjuvant Therapy in HER2-Negative
Breast Cancer with Exploratory Oncotype
DX® Assessments: A Sarah Cannon
Research Institute Phase II Trial
Nancy W. Peacock MD1,2, Denise A. Yardley MD1,2, Carolyn Hendricks MD3, Sang
Yoon Huh MD4, Sarah Ketchum MD5, Calvin Chao MD6, Carl Yoshizawa PhD6, Mythili
Shastry PhD1, Brian Strike MS1, Howard A. Burris MD1,2, John D. Hainsworth MD1,2
1Sarah
Cannon Research Institute, Nashville, TN; 2Tennessee Oncology, PLLC,
Nashville, TN; 3Center for Cancer and Blood Disorders, Bethesda, MD; 4Providence
Medical Group, Terre Haute, IN; 5Mercy Hospital, Portland, ME; 6Genomic Health,
Redwood City, CA
Peacock NW, et al. ASCO 2011. Abstract 1066 (poster presentation).
19
Exploratory Oncotype DX®
Assessments
• The Oncotype DX Breast Cancer Assay uses reverse
transcriptase-polymerase chain reaction (RT-PCR) to assess
a panel of 21 genes in tumor tissue to calculate a score
predictive of the probability of recurrence (Paik et al 2004)
• In this study, Oncotype DX Recurrence Score® results were
calculated from tumor tissue obtained at baseline and after
neoadjuvant therapy in this ER+/ER- patient population
• Recurrence Score results were then correlated with pCR rate.
These correlations may provide preliminary information
regarding the use of the Oncotype DX assay to predict
response to neoadjuvant chemotherapy
Peacock NW, et al. ASCO 2011. Abstract 1066 (poster presentation).
20
Oncotype DX® Assessments (N=38)
Paired Oncotype DX assessments (baseline biopsy and surgical
specimen) were available for 38 patients at the time of the interim analysis.
p=0.025 from logistic regression
Correlation of
Recurrence Score® result
and pCR (38 patients;
7 pCR events)
High
Int.
Low
No
Yes
Pathologic Complete Response
Peacock NW, et al. ASCO 2011. Abstract 1066 (poster presentation).
21
Correlation Between Recurrence Score®
and pCR Rate
• The interim assessment of Oncotype DX® Recurrence Score values
indicated that high Recurrence Scores from the core biopsy may
predict a higher likelihood of pCR with neoadjuvant
ixabepilone/cyclophosphamide.
• Recurrence Score values from the pre-treatment core biopsy and
the post-treatment surgical specimen were similar in most patients
(Spearman r = 0.81, 95% CI: 0.56-0.92)
Peacock NW, et al. ASCO 2011. Abstract 1066 (poster presentation).
22
Patients with a Low Recurrence Score® Result Are Less
Likely to Respond to Neoadjuvant Chemotherapy
Neoadjuvant
Anthracyline-Taxane Treatment
(N=89)
Neoadjuvant
Docetaxel Treatment
(N=72)
P=0.005
P=0.0079
Gianni L, et al. J Clin Oncol. 2005;23(29):7265-7277.
Chang JC, et al. Breast Cancer Res Treat. 2008;108 (2):233-240.
23
The Precision of the Recurrence Score® Result Defines
Individual Biology for ER+ Breast Cancer
Distant recurrence at 10 years
CONTINUOUS BIOLOGY
40%
35%
30%
25%
20%
15%
10%
5%
0%
0
5
10
15
20
25
30
35
40
45
50
Recurrence Score value
LOW RECURRENCE SCORE DISEASE
Indolent
Hormone therapy-sensitive
Minimal, if any, chemotherapy benefit
HIGH RECURRENCE SCORE DISEASE
Aggressive
Less sensitive to hormone therapy
Large chemotherapy benefit
Paik S, et al. N Engl J Med. 2004;351:2817; Paik S, et al. J Clin Oncol. 2006;24:3726; Habel LA, et al. Breast Cancer Res. 2006;8:R25-R39.
Can You Consistently Predict
the Recurrence Score® Result
with One or a Combination of
Traditional Measures?
Evaluation of Recurrence Score® and
Traditional Clinicopathologic
Assessments in a Large Estrogen
Receptor-Positive, Lymph NodeNegative Patient Cohort
Liebermann N,1 Baehner FL,2,3 Soussan-Gutman L, Klang S, Yoshizawa
C,3 Shak S,3 Siegelmann-Danieli N
1Clalit
Health Services, Tel Aviv, Israel; 2University of California, San
Francisco, San Francisco, CA; 3Genomic Health, Inc., Redwood City, CA;
4Teva Pharmaceutical Industries, Netanya, Israel; 5Maccabi Health
Organization, Tel-Aviv, Israel
Liebermann N, et al. ASCO 2011. Abstract 632 (poster presentation).
26
Objectives
• To examine, in a large cohort of ER-positive (ER+),
HER2-negative (HER2-), lymph node-negative (LN-)
patients selected for the Oncotype DX® breast cancer
assay, the distribution of the Recurrence Score® (RS)
result and RS groups, within individual and joint
categories of age, tumor size, and grade.
Liebermann N, et al. ASCO 2011. Abstract 632 (poster presentation).
27
Spearman Correlation Coefficients for
Continuous Recurrence Score® Result
with Individual Prognostic Factors
n
Spearman Correlation
With Continuous RS
95% CI
Age (years)
1864
-0.09
(-0.13, -0.04)
Tumor Size (cm)
1849
0.07
(0.02, 0.12)
Tumor Grade
1530
0.33
(0.28, 0.37)
Characteristic
RS is poorly correlated with age and tumor size, and
modestly correlated with tumor grade.
RS, Recurrence Score
Liebermann N, et al. ASCO 2011. Abstract 632 (poster presentation).
28
Recurrence Score® (RS) Groups
by Age / Tumor Size / Tumor Grade
RS Groups by Patient Age
<50 yrs
(n=367)
RS Groups by Tumor Grade
Grade 1
(n=277)
≥50 yrs
(n=1497)
Grade 2
(n=964)
RS Groups by Tumor Size
Grade 3
(n=289)
≤2 cm
(n=1447)
>2 cm
(n=402)
• Not all grade 1 tumors have low RS values.
• Only 31% of grade 3 tumors have high RS
values.
• Small tumors have proportionately fewer high
RS values.
• However, there is a range of RS values
across both categories of tumor size.
Liebermann N, et al. ASCO 2011. Abstract 632 (poster presentation).
29
Recurrence Score® (RS) Groups
by Age / Tumor Size / Tumor Grade
RS Groups
by Age and Tumor Size
Age <50 yrs,
Size ≤2 cm
(n=302)
Age <50 yrs,
Size >2 cm
(n=64)
• The likelihood of a high RS value in tumors >2 cm is approximately twice
that for tumors ≤2 cm.
• However, there is a range of RS values for each category of age and size.
Age ≥50 yrs,
Size ≤2 cm
(n=1145)
Age ≥50 yrs,
Size >2 cm
(n=338)
RS Groups
by Age and Tumor Grade
Age <50 yrs,
Grade 1 (n=56)
Age <50 yrs,
Grade 2 (n=197)
Age <50 yrs,
Grade 3 (n=66)
Age ≥50 yrs,
Grade 1 (n=221)
Age ≥50 yrs,
Grade 2 (n=767)
RS Groups
by Tumor Size & Tumor Grade
Age ≥50 yrs,
Grade 3 (n=223)
Size ≤2 cm,
Grade 1 (n=247)
Size ≤2 cm,
Grade 2 (n=775)
• There are relatively few high RS values among patients with low grade,
but not all low grade tumors have low Recurrence Scores.
• There is otherwise a wide range of Recurrence Scores among joint
categories of grade and tumor size.
Size ≤2 cm,
Grade 3 (n=194)
Size >2 cm,
Grade 1 (n=29)
Size >2 cm,
Grade 2 (n=183)
Size >2 cm,
Grade 3 (n=37)
Liebermann N, et al. ASCO 2011. Abstract 632 (poster presentation).
30
Summary and Discussion
• There is a range of Recurrence Score® biology
across patient age, tumor size, and tumor grade in
lymph node-negative, ER-positive early stage
breast carcinoma
• The correlations between the Recurrence Score
and these clinicopathologic factors are small for
age and tumor size, and modest for tumor grade
• The Recurrence Score cannot be predicted by
traditional clinicopathologic variables
Liebermann N, et al. ASCO 2011. Abstract 632 (poster presentation).
Practice and Health
Economic Analyses
Evaluating Utilization
Characteristics for the Oncotype DX®
Recurrence Score®
Chen C, Patt DA, Kazzaz D, Shankleton J,
Forsyth M, Ganesh R
Department of Healthcare Informatics,
US Oncology Network, The Woodlands, Texas
Chen C, et al. ASCO 2011. Abstract 625 (poster presentation).
33
Objective
• Primary Objective
– To determine whether practice or patient characteristics
drive Oncotype DX® testing and whether there is an
association between the results of the Oncotype DX test
and the patient’s likelihood of receiving chemotherapy
• Secondary Objectives
– To determine, of breast cancer patients in the US
Oncology Network who are eligible for Oncotype DX
testing, how many received the test
– To determine whether there are patient or practice
characteristics that influence whether patients receive
Oncotype DX testing
Chen C, et al. ASCO 2011. Abstract 625 (poster presentation).
34
Summary and Discussion
• Clinical factors influenced whether patients received
the Oncotype DX® assay in the USO Network
– Oldest patients (>76) had decreased utilization of
Oncotype DX
– Utilization is greatest in patients with tumors on the border
of chemotherapy choice as dictated by pre-Oncotype DX
guidelines (T1b, T1c, T2 tumors)
– Utilization diminished in the very small (T1a) and very
large (T3 & T4) tumors where decisions were likely based
on size, given favorable pathology
• Oncotype DX testing influences decision-making about
whether to incorporate chemotherapy into the care of
patients with early-stage breast cancer
Chen C, et al. ASCO 2011. Abstract 625 (poster presentation).
A Cost-Benefit Analysis of the
21-Gene Breast Cancer Assay within a
Canadian Health Care System
S Hassan1, N Mittmann1,2
1
Health Outcomes & PharmacoEconomics (HOPE) Research
Centre, Division of Clinical Pharmacology, Department of
Medicine, Sunnybrook Health Sciences Centre, Toronto
2
Department of Pharmacology, University of Toronto, Toronto
Hassan S and Mittmann N. ASCO 2011. Abstract 6111 (poster presentation).
36
Cost Benefit Analysis of the 21-Gene Breast Cancer
Assay within a Canadian Health Care System
• Objective
– To quantify budget impact relative to the addition of the 21BCA to a
Canadian provincial framework
• Conclusion
– In this analysis model, all patients with low Recurrence Score® (RS)
results and 50% of intermediate RSs avoided chemotherapy
– The budget impact of the cost of the 21BCA would be offset by a
reduction in costs associated with treatment. The use of the 21BCA in
BC patients using this model would have resulted in more than $11M
(CDN) per year in savings to the health care system
– This conservative analysis may show even greater costs savings if
adverse event costs were considered
Hassan S and Mittmann N. ASCO 2011. Abstract 6111 (poster presentation).
Validation of a 12-Gene Colon
Cancer Recurrence Score® in
Stage II Colon Cancer Patients
from CALGB 9581
A.P. Venook1, D. Niedzwiecki2, M. Lopatin3, M. Lee3, P. N.
Friedman4, W. Frankel5, K. Clark-Langone3, C. Yoshizawa3, C.
Millward3, S. Shak3, R. M. Goldberg6, N. N. Mahmoud7, R. L.
Schilsky4, M. M. Bertagnolli8
1. University of California, San Francisco, San Francisco, CA; 2. Duke University, Durham, NC; 3.
Genomic Health, Redwood City, CA; 4. The University of Chicago, Chicago, IL; 5. The Ohio State
University, Columbus, OH; 6. University of North Carolina at Chapel Hill, Chapel Hill, NC; 7.
University of Pennsylvania, Philadelphia, PA; 8. Brigham and Women's Hospital, Boston, MA
Venook AP, et al. ASCO 2011. Abstract 3518 (poster presentation).
38
CALGB 9581: A Unique Opportunity to Test the
Oncotype DX® Assay in Low/Standard Risk
Stage II Colon Cancer
• Sampling design with 690
evaluable patients,
including 162 recurrence
events, representative of
1,672 stage II colon
cancer patients from
CALGB 9581
• A population where risk
discrimination is
challenging with
conventional clinical and
pathologic factors
Venook AP, et al. ASCO 2011. Abstract 3518 (poster presentation).
Patient Characteristics of the
Validation Studies
Characteristic
|
CALGB
|
QUASAR
39
CALGB 9581: Second Successful
Prospectively-Designed Confirmation Study
Relationship Between RS and Risk of
Recurrence at 5 Years
RS, Recurrence Score®
Venook AP, et al. ASCO 2011. Abstract 3518 (poster presentation).
40
CALGB 9581: Improving the Ability to
Discriminate High vs. Low Risk of Recurrence
in Standard Risk Stage II Colon Cancer
Ability to Discriminate High vs
Low Risk of Recurrence:
RS Group
% of
Patients
RS, Recurrence Score®
Venook AP, et al. ASCO 2011. Abstract 3518 (poster presentation).
Average 5-Year
Recurrence Risk for T3
MMR-P Patients (95% Cl)
Quantitative Gene Expression in Primary
and Highest Gleason Patterns Identifies
Genes Associated with Clinical
Recurrence in Prostate Cancer After
Radical Prostatectomy
Eric A Klein1, Sara M Falzarano2, Tara Maddala3, Carl Millward3,
William F Novotny3, Robert J Pelham3, Diana Cherbavaz 3, Cristina
Magi-Galluzzi2
Cleveland Clinic 1Glickman Urological and Kidney Institute and 2Pathology and
Laboratory Medicine Institute, Cleveland Clinic
3 Genomic Health, Inc., Redwood City, CA
Klein EA, et al. ASCO 2011. Abstract 4663 (poster presentation).
42
Prostate Cancer Is Widely Over-treated to
Benefit Very Few
Need Is Clear
• >750,000 biopsies1; 200,000 cancers diagnosed2 (U.S.
2009)
• Vast majority get aggressive surgery/radiation therapy;
<10% patients choose active surveillance
1
Marks et al, Urology 2007; 69:532.
2
Jemel et al, CA Cancer J Clin 2009;59: 225-249.
43
Which Men Have Clinically Significant
Prostate Cancer Requiring Treatment?
• Surgery and radiation therapy are common
treatments for prostate cancer
• However
– Between 19 and 48 men have to be treated to prevent
one death from prostate cancer
– Over 50% of men suffer incontinence and/or
impotence after surgery
Schröder FH et al, N Engl J Med. 2009; 360:1320-1328.
44
Unique Challenges in Prostate Cancer
• Small tumors common and biopsy tissue very
small; required modification of our RT-PCR
platform for lower input RNA
• Multi-focal disease common; sampling with
standard biopsy procedures may confound
accurate prediction of risk
45
Study Objectives
• Overall Objective
– Development of a biopsy-based genomic assay to
discriminate indolent from aggressive prostate cancer
• Objective of Current Study:
– Quantitative assay of gene expression in radical
prostatectomy (RP) tissue
• Identify genes whose expression predicts clinical recurrence
after RP
• Assess the ability of gene expression to predict clinical
recurrence (cR) in the context of Gleason grade
heterogeneity
• Assess the ability of gene expression to provide recurrence
risk information beyond that provided by Gleason grade
Klein EA, et al. ASCO 2011. Abstract 4663 (poster presentation).
46
Patients, Samples, Methods, and Genes
2,600 Patients with RP 1987-2004
501 Patients Selected
127 Clinical Recurrence
374 Disease Free
Quantitative RT-PCR using
Oncotype DX® RT-PCR
technology for 738 Genes
6 Reference Genes
732 Cancer-Related Genes
51 Excluded for
insufficient tumor
7 Did not meet inclusion
criteria
2 Outlier gene profile
441 Patients Evaluable for
Clinical /Pathology/Gene
Expression
Klein EA, et al. ASCO 2011. Abstract 4663 (poster presentation).
5 Cancer genes
excluded
727 Evaluable Cancer
Genes
47
Methods
Specimens re-reviewed centrally for assignment of stage & grade
(2005 Consensus)
Manual microdissection
Two tumor foci separately
analyzed
Primary Gleason pattern
Highest Gleason pattern
RNA extraction
Gleason
Pattern 3
Gleason
Pattern 4
RT- PCR for each gene
Data analysis
Identify specific gene expression levels associated with outcome (cR, bR, PCSS)
Cox regression controlling the false discovery rate at 10%
Klein EA, et al. ASCO 2011. Abstract 4663 (poster presentation).
48
Genes are Associated with Clinical Recurrence in Both
Sample
No. of Genes
Associated
with Clinical
Recurrence
Primary
Gleason Pattern
295
Highest
Gleason Pattern
297
Both Primary and
Highest Gleason Pattern
235
HR for association of gene expression with cRFI in HGP
the Primary and Highest Gleason Pattern Samples
2.4
1.6
1.0
0.7
0.4
0.4
0.7
1.0
1.6
2.4
HR for association of gene expression with cRFI in PGP
Genes significantly associated with cRFI in both PGP and HGP
Genes significantly associated with cRFI in PGP only
Genes significantly associated with cRFI in HGP only
Genes not significantly associated with cRFI in PGP or HGP
\\Lims\medical\Prostate\CCFProstateGI(09-002)\Analysis\USCAP_2011\Programs\g_ScatterStdHRcRFI_PGPvHGP.sas
Klein EA, et al. ASCO 2011. Abstract 4663 (poster presentation).
12JAN2011:1459 SCL:00.05.96 tmaddala
49
Genes Associated With Clinical Outcome
Cluster Into Co-Expression Groups with
Related Biological Functions
Klein EA, et al. ASCO 2011. Abstract 4663 (poster presentation).
50
Hazard Ratios for Genes Associated with
Clinical Recurrence in Both Primary and
Highest Gleason Pattern Samples
Univariate Std. HRs for cRFI
Cox PH Regression
Worse Outcome
Better Outcome
Stress
DUSP1
FOS
Migration
GSTM2
PPAP2B
Proliferation
BIRC5
TOP2A
ECM
COL3A1
FAP
0.3
0.6
1.0
1.8
Std. Hazard Ratio for Clinical Recurrence
Stnd Hazard Ratio
for ClinicalPrimary
Recurrence
Highest
\\L i m s \m e d i c a l \Pro s ta te \CCFPro s ta te GI(0 9 -0 0 2 )\An a l y s i s \AAGUS\Pro g ra m s \g _ Fo re s tc RFIStd HR4 Ge n e Grp _ x GP.s a s
Klein EA, et al. ASCO 2011. Abstract 4663 (poster presentation).
3 0 M AR2 0 1 1 :1 2 2 6 SCL :0 0 .0 6 .0 5 j y u e
3.2
51
Each Gene Group Add Prognostic Value
Beyond Clinical and Pathological Covariates
Worse Outcome
Better Outcome
DUSP1
Stress
FOS
Unadjusted
GSTM2
Adj. for path Tstage
PPAP2B
Adj. for surgical GS
BIRC5
Adj. for Gleason
pattern
TOP2A
Adj. for CAPRA risk
group
COL3A1
Adj. for AUA risk
group
Migration
Proliferation
ECM
FAP
0.3
0.6
1.0
1.8
Std.
Hazard
Ratio
(95% CI)
Std.
HR
(95%
CI)
AUA
Klein EA, et al. ASCO 2011. Abstract 4663 (poster presentation).
GP
CAPRA
GS
3.2
52
Gene Expression Predicts for All
Clinically Important Endpoints
Better Outcome
Worse Outcome
DUSP1
Stress
FOS
cRFI
GSTM2
Migration
bRFI
PPAP2B
Upgrade/
Upstage
BIRC5
Proliferation
PCSS
TOP2A
ECM
COL3A1
FAP
0.3
0.6
1.0
1.8
Std. Relative Risk Estimate (95% CI) (log scale)
PCSS
Klein EA, et al. ASCO 2011. Abstract 4663 (poster presentation).
UpRsk
3.2
Final Thoughts
54
Genomic Health’s Research is Focused
on Optimizing Cancer Treatment
Breast
Renal
Colon
Prostate
55
Closing Thoughts
• Importance of understanding and treating the underlying
individual tumor biology
• Importance of standardizing emerging “next generation”
technologies
• Genomic assays for clinical decision-making must be
“Fit for purpose”
– Clinically validated in prospectively-designed studies
of sufficient size and statistical power
– Supported by evidence in target patient population,
with demonstrated value beyond existing measures
– Standardized and reproducible
– Practical and clinically impactful