1 OncoVue ® Breast Cancer Risk Assessment Test Rita Murry Progressive Medical Enterprises Empowerment through intelligent medicine
1
OncoVue® Breast Cancer Risk
Assessment Test
Rita MurryProgressive Medical EnterprisesEmpowerment through intelligent medicine
Research Background
• Research began in 1989 at Samuel Roberts Noble Foundation on inhibitors of cell growth
• Moved to the Oklahoma Medical Research Foundation in 1993
• Company spun-out in 1999
• Over 20,000 patient specimens analyzed in our genetic risk assessment studies
Personalized Medicine
• Cancer risk using genetics and clinical/personal history
• Cancer as a complex disease with multiple contributing genes
• Identification of cancer prevention drugs
• Cancer therapy targeted to genetically identifiable tumors
Overview
• Background • Model Development
– Study Population
– Genetic Polymorphisms
• Model Description• Validation and Clinical Utility• Testing Process• Case Study
BC is a Complex Disease
• Multiple pathways OMIM (>500 hits)• Excess estrogen and/or progesterone
(length of menstrual life, age of FLB, HRT, obesity, tall stature, etc.)
• Age
Traditional Risk Assessment
• Gail model (1989)– Age interval and lifetime risk
– AGEMEN, AGEFLB, NUMREL, NBIOP
– Recalibrated for prevention trials (1999)
• Claus model (1994) – Family history/age of onset in relatives
Genetic Risk Assessment
• BRACAnalysis® (1995)
– Familial breast and ovarian cancer
• Candidate Genes (1999)
• OncoVue® (2006)
– Multivariate Logistic Regression Model
– Candidate Genes and Personal Factors
• GWAS Markers (2007)
– 7 major SNPs/now a commercial test
Why a New Risk Model?
• Improve individualized breast cancer risk estimation by integrating genetic variation and personal factors
• Single Nucleotide Polymorphisms (SNPs)
• Personal History Measures (PHMs)• Risk estimation compared to the Gail
Model
Genotyping Strategy
Candidate genes vs. GWAS• Efficiently identify risk associations
• Minimize false discovery rate/misclassification of variants
• Functional polymorphisms likely to be causative polymorphism (not just in LD)
Gene Selection Criteria
• Biological/physiological relevance to cancer
• Known or predicted functional consequences
• Amino acid charge (coding regions), promoters (transcription), splice junctions/3'UTR (mRNA half life or translational control)
• Minor allele common across ethnicities (range 1-50%; median 30%; mean 28%)
125 SNPs- Many Pathways
• Steroid Hormone Receptors/Metabolism (25) • DNA Damage/Repair (30)• Xenobiotic/Conjugation/Detoxification (15)• Cell Cycle and Apoptosis (15)• Growth Factors and Signaling (12)• Immune Modulation/Cytokines (9)• Invasion/Metastasis (8)• Lipid Metabolism (5)• Free Radical Scavengers (3)• Angiogenesis (3)
Genotyping Platform
• Allele Specific Primer Extension (ASPE)- FlexMap beads
• Luminex benchtop flow cytometer• Multiplexed into 6 reactions• Double-strand sequence verified• < 5 ng DNA/multiplex• Genotyped over 20,000 individuals in
R&D
Case-Control Study
• OK, WA, CA, KS, FL, SC• Mammography screening clinics• Community breast cancer awareness
events• Enrolled unrelated individuals• Completed questionnaire on personal
medical history, family history of cancer and lifestyle
• Buccal cells in commercial mouthwash for DNA isolation
Study Population
• Demographic variables and allelic frequencies were similar across:
– Region of enrollment
– Clinic vs. Community
Model Building Strategy
• Combined all Caucasian women in study
• Assigned with a pseudo-random number generator into:
– Model Building (80%)
– Validation Set 1 (20%)
• Validation Set 2 African American women from same catchment areas
Sample Sets
African American Women Ages 30-69
Caucasian Women
Ages 30-69
Validation Set 1
Cases = 400
Controls = 793
Total = 1193
Valid Set 2
Cases = 164Controls = 417
Total = 581
Model Building
Cases = 1671
Controls = 3351
Total = 5022
Model Building Strategy
• Checked HWE in controls (117 passed)
• Systematically evaluate SNPs and PHM associations with case-control status
– Term individually
– Term*Term Interaction
– Term*Age Interaction
CYP11B2
• Key enzyme that ultimately converts 11-deoxycorticosterone to aldosterone
• C/C genotype associated with increased risk of type II diabetes
• Type II diabetes associated with increased risk of BC in postmenopausal women
CYP11B2 Age-Specific
0.5
1
1.5
2
30 40 50 60 70
Age (years)
OR
C/C (974)
C/T (2455)
Ref
R2 = 0.95 (35-65)
R2 = 0.89 (35-65)
11-deoxycorticosterone to aldosterone, Type II Diabetes
Ralph, D. et al. CANCER, 2007, 1940-48
Final Terms
4
Model Building Steps
1Univariate 2 p-value took
top 25%
2Forward stepwise selection modeling
3
p-value to enter ≤ 0.1
p-value to retain ≤0.05
Bootstrap 5000X for SE
Strategy for Iterative Analyses
Performed individually on:
• Entire dataset Ages 30-69– Winners incorporated into model
• Stratified Ages 30-49, 50-69
– Winners incorporated into model
• Stratified by PHMs (First Degree Relative Status)
OncoVue®
SNPs SNP*AGE PERSONAL HISTORY MEASURES
30 - 69
SNP*AGE PERSONAL HISTORY
MEASURES*AGE
30 - 49 w/o
FDR
SNP*AGESNPs30 - 49
w/1 FDR
30 50 60
49
70
69
40
Total of 22 SNPs in 19 Genes and 5 PHMs
All Ages
Term Age Interaction
Individual Terms
CYP1A1
ACACA (IVS17)
IGF2
NUMREL
ACACA (5’UTR)
VDR
XRCC2
AGEFLB
NBIOP
MSH6
CYP11B2
ESR1
ERCC5
Genes, SNPs and Function
BRCA1 Interaction
• ACACA (IVS17)
• ACACA (5’UTR)
• ACACA (PIII)
DNA Repair• MSH6
• RAD51L3
• XPC
• ERCC5
• XRCC2
Steroid Hormone Metabolism
• COMT
• CYP11B2
• CYP19
• CYP1A1
• CYP1B1 (N453S)
• CYP1B1 (R48G)
• ESR1
• VDR
Genes, SNPs and Function
Cell Cycle/Apoptosis
• KLK10
• TNFSF6
Detoxification
• EPHX
• SOD2
• INS
• IGF2
Growth Factors
Risk Model Performance
• Evaluate with informative measures that reflect improvement in individual risk estimation
• Not a traditional diagnostic test
• Gail Model is current clinical tool for risk estimation for the majority of women
Positive Likelihood Ratio (PLR)
PLR = Elevated Cases/All Cases Elevated Controls/All Controls
• Not sensitive to population characteristics or disease prevalence
• Completely Random = 1.0• Increase represents improvement• Elevated risk threshold ≥12%• 1.5X SEER average risk for ages 30-69
OncoVue® Performance
• Model Building (1671 Ca/3351 Co)• Validation 1 (400 Ca/793 Co)• Validation 2 (164 Ca/417 Co)• External Blinded Validation (169 Ca/177 Co)
– Marin County California 1997-1999
– Removed Ca/Co status
– Provided buccal cells and PHM
– Genotyped and scored for OncoVue
– Returned for evaluation by UCSF
Fold Improvement-12% Threshold
PLR Sample Set
OncoVue® Gail Model
Fold Improvement (95% CI)
p-value
Model Building 2.1 1.2 1.8 (1.4, 2.2) <0.0001
Validation 1 2.4 1.5 1.7 (1.1, 2.5) 0.024
Validation 2 3.2 1.4 2.2 (1.1, 5.3) 0.034
Marin County 2.2 0.90 2.4 (1.1, 5.6) 0.036 *PLR = Positive likelihood ratio, CI = Confidence Interval
Fold Improvement
• Trend in fold improvement increases at higher cut-off thresholds
• At 20% the fold improvement
– Model Building= 3.0 (p<0.0001)
– Validation Set 1 = 2.1 (p=0.07)
– Validation Set 2 = no controls >20%
PLR for BRCA1
For ages 30-69 - assuming a RR=8.0 for BRCA1 carrier and 8% SEER average risk
Cases Controls
BRCA+ 410 590 = 1000
BRCA - 80 920 = 1000
490 1510
PLR = 410/490 ÷ 590/1510 = 2.1 (1.9, 2.3)
Increase in Cases Identified at Elevated Risk
Gail Model OncoVue®
Sample Set Cases Controls Cases Controls
No. of Additional Detected
Cases
Percent more Cases
over Gail
Model Building 454 760 577 760 123 27%
Validation 1 118 161 135 161 17 14%
Validation 2 32 56 42 56 10 31%
Marin County 37 43 56 43 19 51%
Blinded Validation
• OncoVue showed a 2.4-fold statistically significant (p=0.036) improvement over the GM alone
• Additional 51% of cases accurately assigned elevated risk
• GM underestimated risk for those individuals actually at highest risk of developing breast cancer
OncoVue Summary
• First DNA-based test for estimating age-specific risk of developing sporadic breast cancer
– Applicable to the majority of women– Not a test for familial breast cancer risk (Myriad)
• Single integrated statistical model or algorithm• Developed in a model building set of 5000+
cases/controls and initially validated in two independent populations1
• Subsequently have completed a blinded validation in high risk population from Marin County, CA2
1 Jupe, E. et al. Proc. AACR 2008; 49: 451.
2 Dalessandri, K. et al. Presentation # 502, San Antonio Breast Cancer Symposium, December 2008.
OncoVue Summary
• Outperforms other genetics-based testing for sporadic breast cancer that were developed using different strategies1,2
• Presented at San Antonio Breast Cancer Symposium in 2008 and was subject of got widespread AP press release and press conference
• Paper chosen for late-breaking presentation at 2009 SABCS
1 Jupe, E. et al. Presentation # 3177, San Antonio Breast Cancer Symposium, December 2009.
2 Dalessandri, K. et al. Presentation # 3057, San Antonio Breast Cancer Symposium, December 2009.
Current Clinical Utilization
• 33 Comprehensive Breast Care Centers in 24 states (Breast surgeons, Oncologists, Mammography)
• Important tool to drive decision making and stratify risk and is applicable to most patients
• Over 2,600 clinical test results delivered
• Risk perception and behavior studies found that “knowledge is power”
• Reimbursement under existing CPT Codes
Testing Process is Simple
Clinical Decision Tool
• Decision support tool to guide screening and prevention options
• Identification of women that are candidates for preventative anti-estrogen therapies (tamoxifen, raloxifene) following guidelines from American Society of Clinical Oncology (ASCO)
• Identification of women who may benefit from supplemental MRI screening following guidelines from the American Cancer Society (ACS)
Other Clinical Utility
• Decisions regarding the use of hormone replacement therapy (HRT) in post-menopausal patients
• Potential value in identifying higher risk patients that need more frequent screening than the recent controversial recommendations from the US Preventive Services Task Force (USPSTF)
• Although it is not a diagnostic test for disease – high risk score has led to additional testing that brings about early diagnosis
OncoVue Risk and Early Diagnosis
• 50 year old woman presented for routine screening• Mammogram appeared normal• High Oncovue risk with 5 year = 4.9• More sophisticated imaging found 2mm tumor
Changing Health Care Model
Diagnosis Detection Treatment
Monitor Therapy
Prevention or Earliest Possible Detection
Genetic Predisposition Testing
Intervention
Outcome?
Change Treatment
Work in Progress• The Next Generation of OncoVue
– SBIR Phase I (miRNA polymorphisms)– OCAST-OARS (miRNA targets)– OCAST-OARS (Copy Number Variation)– AVON Foundation – Collaboration with Marin
County Health Department (breast density)
• Risk tests for other cancers – Ovarian
Acknowledgments
InterGenetics• Sharmila Manjeshwar, PhD• Daniele DeFreese, MS• Bobby Gramling, MS• Thomas Pugh, MS• Laura Blaylock, BS
Statistical Consultants• Christopher Aston, PhD• Dr. Lue Ping Zhou, PhD• Nicholas Knowlton, MS
OUHSC• John Mulvihill, MD
University of California San Francisco• Kathie Dalessandri, MD• Margaret R. Wrensch, PhD • John K. Wiencke, PhD• Dr. Rei Miike, PhD
Buck Institute for Age Research
• Christopher C. Benz, MD
Zero Breast Cancer• Georgianna Farren, MD
Acknowledgments
Marin County Health Department• Mark Powell, MD, MPH• Lee Ann Prebil, PhD • Rochelle Ereman, MS, MPH
Acknowledgments
Funding Sources• NIH – SBIR
• US Army BCRP
• Oklahoma Center for the Advancement of Science and Technology
• American Cancer Society
• Swisher Family Trust
• Presbyterian Health Foundation
• Oklahoma Life Sciences Fund
Clinical and Scientific Advisory Board
• John Mulvihill, M.D.– Professor of Pediatrics; Chief of the Human Genetics Section;
Kimberley V. Talley Chair, Children’s Medical Research Institute, University of Oklahoma Health Sciences Center
• Debra Mitchell, M.D.– Medical Director, Breast Imaging of Oklahoma; Former Imaging
Director, University of Oklahoma Breast Institute; Adjunct Associate Professor, Dept. Radiological Sciences, University of Oklahoma Health Sciences Center
• Christopher Aston, Ph.D. – Biostatistics/Bioinformatics Core Director, General Clinical
Research Center; Associate Professor, Dept. Pediatrics and Genetics, University of Oklahoma Health Sciences Center
• Linda Thompson, Ph.D.– Member, Immunobiology and Cancer Program, Oklahoma Medical
Research Foundation; Adjunct Professor, Department of Microbiology and Immunology, University of Oklahoma Health Science Center
Academic Collaborators and Research Participants