Download 2 - Gynecoland

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts
no text concepts found
Transcript
®
OncoVue
Breast
Cancer Risk
Assessment Model
Dr. Eldon Jupe
Vice President
Clinical Laboratory Director
March 17, 2012
Disclosure
• Co-founder of InterGenetics Incorporated
• Employee with Stock Options
• Vice President and Director of CLIA Laboratory
®
OncoVue
Background and
Research History
Breast Cancer Risk
• Complex Disease
• Age
• Excess estrogen and/or progesterone
− Length of Menstrual Life
− Age of First Live Birth
− Hormone Replacement Therapy
− Obesity, Tall Stature, etc.
Breast Cancer Risk Assessment
• Gail Model (1989) also called NCI-BCRAT
– Age interval (5-Year) and Lifetime Risk
– AGEMEN, AGEFLB, NUMREL, NBIOP
– Recalibrated in prevention trials (1999)
• Claus Model (1994)
– Family history/age of onset in relatives
• BRACAnalysis® (1995)
– Familial breast and ovarian cancer
– Gene testing for mutations
Familial Breast Cancer Genetics
• Families with strong history
• Multiple first degree relatives with BC
• Woman with breast and ovarian cancer
• Male breast cancer in family
• Hereditary predisposition (5-10% of cases)
• Approximately half with gene mutations
• Mostly BRCA1, BRCA2 (1994)
• PTEN (Cowden Syndrome)
• TP53, CHEK2 (Li-Fraumeni Syndrome)
Familial Breast Cancer Genetics
• Hope was that familial cancer genes would
be widely applicable
• Now almost 20 years later
• No additional familial genes found for half of
inherited breast cancers
• No association of BRCA1/2 with sporadic
breast cancer
Sporadic Breast Cancer Genetics
• Several twin studies indicate cancer, in
general, and breast cancer, in particular,
has strong genetic component
• Can the influence of many common genetic
variants account for most breast
cancers?
• Single Nucleotide Polymorphisms (SNPs)
SNPs
• GENOME: The total DNA
content of all genes on the
chromosomes of an individual
• GENOMICS: the study of
genes and their function
• 3 billion base pairs in the
human genome
• All individuals are 99.9%
identical in sequence
• SNP = Single base pair
changes leading to our
inherited differences and
susceptibility to disease
Research Background
• Oklahoma Medical Research Foundation (1993)
• Searching for inhibitors of cell division
• First identified in regenerating rat liver
• Then found human gene
• Tumor suppressor gene (1995)
• Identified a human SNP with low activity (1996)
• Does not suppress human tumor formation
• High activity vs. low activity of biological product
SNP Influences Tumor Growth
T-Form (Grows Rapidly)
C-Form (Growth Suppressed)
SNP Influences Tumor Growth
T-Form (Ki-67-positive cells stained)
C-Form (50% Reduction in positives)
Genomics Applied to Cancer
• Hypothesis that this SNP has a role in
influencing breast cancer susceptibility
• Funded to perform breast cancer casecontrol study
• Used microarrays to identify genes
differentially expressed tumor vs. normal
• Expanded studies from single gene tumor
suppressor characterization to multigenic
Development of
®
OncoVue Breast
Cancer Risk
Assessment Model
What is the Need?
• Most breast cancers sporadic (90-95%)
• Gail model (clinical factors)
• Discriminatory accuracy is modest
r = 0.58 for Nurse’s Health Study
r = 0.50 is random
• Claus model (family history w/age of
onset)
• Indirect measures of genetics
• Directly determine the genetics
Goal
To develop a significantly improved
risk assessment model for “sporadic”
breast cancer utilizing genetic and
clinical/personal risk factors.
Our Approach
• Cancer risk using genetics/clinical history
• Cancer as a complex disease with multiple
contributing factors
• Over 20,000 patient specimens analyzed in
development and validation studies
• Application for identification of cancer
prevention drugs
Candidate Gene Markers
• Thousands considered
− Published papers/our research
− Reviews
− Meta-analyses
− Genomic databases
− OMIM, CGAP, etc.
• Ultimately selected ~125 SNPs
• Pathways influencing breast/other cancers
Selection Criteria
• Gene variations causing functional changes
• Coding or regulatory regions
− Protein coding regions (aa change)
− Promoter/mRNA transcription
− Intron or 3′-UTR/mRNA stability
• Published associations ↑ or ↓ risk
• Breast and/or other cancers
• Minor alleles common in major ethnic groups
− Range 1-50%, Mean 28%, Median 30%
125 SNPs – Many Pathways
• Steroid Hormone Receptors/Metabolism
• DNA Damage/Repair
• Xenobiotic/Conjugation/Detoxification
• Cell Cycle and Apoptosis
• Growth Factors and Signaling
• Immune Modulation/Cytokines
• Invasion/Metastasis
• Lipid Metabolism
• Free Radical Scavengers
• Angiogenesis
Case-Control Study
• Enrolled in 6 distinct regions of US
• Buccal cells collected in mouthwash
• DNA isolated and genotyped
− 125 candidate SNPs
• Extensive questionnaire
− Personal, clinical and family history
Model Building Strategy
• Multivariate logistic regression modeling
• SNPs (117) and traditional risk factors (50)
• Interactions and age-specific associations
• 1,671 cases/3,351 cancer-free controls
• White participants for model building
• Evaluate associations with Ca-Co status
– Term Individually
– Term*Term Interaction
– Term*Age Interaction
Age Influences Genetics
2−
C/C (974)
C/T (2455)
R2 = 0.95
(35-65)
R2 = 0.89
(35-65)
1.5−
OR
Ref
1
0.5
_______________________________________________
30
40
50
60
70
Age (years )
CYP11B2 Age-Specificity (CANCER 109: 1940-1948, 2007)
Model Building Steps
3
Bootstrap 5000X for SE
1
Univariate 2
p-value took
top 25%
2
Forward
stepwise
selection
modeling
p-value to
Enter ≤ 0.1
p-value to
Retain ≤ 0.05
 22 SNPs in 19 Genes
 Age, Age FLB, FDR with BC, Biopsies/Outcome
Final
Terms
4
Genes, SNPs and Pathways
Steroid Hormone Metabolism
• COMT
• CYP1B1 (N453S)
DNA Repair
• MSH6
• CYP11B2 • CYP1B1 (R48G)
• RAD51L3
• CYP19
• ESR1
• XPC
• CYP1A1
• VDR
• ERCC5
• XRCC2
BRCA1 Interaction
• ACACA (IVS17)
• ACACA (5’UTR)
• ACACA (PIII)
Genes, SNPs and Pathways
Cell Cycle/
Apoptosis
Growth
Factors
• KLK10
• INS
• TNFSF6
• IGF2
Detoxification
• EPHX
• SOD2
Performance of
®
OncoVue Breast
Cancer Risk
Assessment Model
OncoVue® Performance
• Comparisons to the BCRAT
− Risk factor based/widely used
• Training set (1,672 Ca/ 3351 Co)
• Independent Validation Sets
− 376 Cases/761 Controls (White)
− 149 Cases/346 Controls (Black)
− 79 Cases/233 Controls (Hispanics)
• Women ≥ 35 years of age
OncoVue® Performance
• Odds Ratio (OR) at different risk thresholds
− 5-year (≥ 1.5, 2.0, 2.5, ...)
− Lifetime (≥ 12, 13, 14,…)
• Counts of Cases and Controls above and
below thresholds
• OR = # of Cases Above/ # of Controls Above
# of Cases Below/ # of Controls Below
• Odds of breast cancer observed at risk level
OncoVue® Performance
• Expectations
• Accurate Risk Estimation
− increasing odds of breast cancer with
increasing risk output
• Random Risk Estimation
− Odds = 1.0
White Participants – 5 Year Risk
Training
8
OncoVue
7
Odds R atio (OR )
6
5
4
3
2
____________________
1
Gail model
Random
0
1
2
3
4
5
6
7
8
Validation
Model R is k S core (% Abs olute R is k)
8
7
O dds R atio (O R )
0
OncoVue
6
5
4
3
2
__________________
1
Gail model
Random
0
0
1
2
3
4
5
6
Model R is k S c ore (% A bs olute R is k)
7
8
White Participants - Lifetime Risk
8
OncoVue
7
Training
5
4
3
Gail model
2
___________________________
1
Random
0
10
12
14
16
18
20
22
24
26
Model R is k S core (% Abs olute R is k )
Validation
8
7
6
Odds R a tio (OR )
Odds R a tio (OR )
6
5
OncoVue
4
3
Gail model
2
_____________________________ Random
1
0
10
12
14
16
18
20
22
Model R is k S core (% Abs olute R is k)
24
26
Black Participants - Lifetime Risk
149 Cases/346 Controls
6
5
Odds Ratio (OR) .
OncoVue
4
3
2
Gail model
1
Random
0
10
12
14
16
18
20
Model Risk Score (% Absolute Risk)
22
24
26
Hispanic Participants - Lifetime Risk
79 Cases/233 Controls
10
OncoVue
9
8
PFRM
Odds Ratio (OR) .
7
6
5
4
3
BCRAT
2
Gail model
1
Random
0
10
12
14
16
18
20
Model Risk Score (% Absolute Risk)
22
24
26
Fold Improvement in ORs at Selected Risk Thresholds
Risk Threshold
5-Year (%)
OR (95% CI)*
®
Fold
Improvement
(95% CI)
p-value
1.3 (1.1, 1.4)
0.002
OncoVue
Gail model
≥ 1.67
1.6 (1.4, 1.9)
1.2 (1.1, 1.3)
≥ 2.0
1.8 (1.6, 2.1)
1.3 (1.1, 1.5)
1.4 (1.1, 1.6)
0.002
≥ 3.0
2.8 (2.2, 3.5)
1.4 (1.2, 1.7)
1.9 (1.4, 2.5)
< 0.0001
≥ 4.0
4.3 (3.0, 6.0)
1.5 (1.2, 2.0)
2.7 (1.8, 4.1)
< 0.0001
Lifetime (%)
Training White ≥ 20
3.6 (2.7, 4.8)
1.7 (1.4, 2.0)
2.1 (1.6, 2.8)
< 0.0001
Validation White ≥20
4.3 (2.2, 8.7)
1.9 (1.3, 2.8)
2.3 (1.1, 4.6)
0.019
Black ≥ 20
4.1 (1.3, 13.8)
1.8 (0.8, 3.9)
2.3 (0.7, 7.7)
0.19
Hispanic ≥ 20
7.0 (1.6, 45.5)
1.0 (0.4, 2.4)
7.0 (1.5, 33.7)
0.015
*OR = Odds Ratio, CI = Confidence Interval
How is This Possible?
• True causative polymorphisms
• Vary in frequency in different ethnic groups
but must strongly influence BC risk
• Genetics is weighted in the presence of
clinical risk factors
• Similar results in Asian- Pacific Islanders
-
Summary
• Validated in two independent populations1
• Blinded validation in high risk population
---Marin County, CA2
• Outperforms other genetics-based testing ----for sporadic breast cancer from GWAS3,4
______
______
______
1. Jupe, E. et al. Proc. AACR 2008; 49: 451.
2. Dalessandri, K. et al. Presentation # 502, San Antonio Breast Cancer -------Symposium, December 2008.
3. Jupe, E. et al. Presentation # 3177, San Antonio Breast Cancer Symposium,
December 2009.
4. Dalessandri, K. et al. Presentation # 3057, San Antonio Breast Cancer
Symposium, December 2009.
Summary
• First DNA-based test for accurately
estimating risk of sporadic breast cancer
• Applicable to the majority of women
• Single integrated statistical model/algorithm
• Demonstrated significant improvement over
current clinical standard
• Decision tool for prevention and screening
options
®
OncoVue
Clinic
in the
Example of Risk Stratification
• Three Women
- Age 44, One First Degree Relative,
- No Biopsies, No Children
- Age Menarche = 15
• Gail model – all have 1.34% 5-Year Risk
• OncoVue – stratifies 3.17%, 1.22%, 0.51%
- Average 5-year risk woman age 44 ~1.0%
- 3.17% diagnosed at age 44
- Genotype ∆ 3.17% vs. 0.51% = 14/22 (64%)
Example of Risk Stratification
• Clinically homogenous group (n=22)
- Ages 42 to 44
- One First Degree Relative
- No biopsies
- No children
Risk Stratification in a Clinically Homogenous Group*
5-Year Risk (%)
Fold Difference
Analysis
OncoVue®
Gail Model
OncoVue®
Gail Model
Range
0.21 − 3.35
1.14 − 1.47
16
1.3
Range (Quintile)
0.55 − 1.67
1.25 − 1.47
3
1.2
Mean Risk Cases (n=8)
1.84
1.40
−
−
Mean Risk Controls (n=14)
0.79
1.35
1.84/0.79 = 2.32 1.40/1.35 =1.0
* Ages 42-46, no biopsies or children, one first degree relative with breast cancer (n=22)
Testing Process is Simple
Clinical Decision Tool
• Guide screening and prevention decisions
• Higher risk patients for more frequent
screening – every 6 months
• Imaging in addition to mammography
- MRI, ultrasound, tomography
•Identification of women that are candidates
for preventative anti-estrogen therapies
• Tamoxifen or Raloxifene
• Decisions about use of hormone
replacement therapy (HRT) in postmenopausal patients
Driver for Early Diagnosis
• Although not a diagnostic test for cancer
high risk score has led to additional
testing that identified tumor
OncoVue Risk and Early Diagnosis
• 50 year old woman presented for routine screening
• Mammogram appeared normal
• Absolute Risk ~ 9.0/Relative Risk ~ 4.5
• More sophisticated imaging identified 2mm tumor
Changing Health Care Model
Intervention
Genetic
Predisposition
Testing
Diagnosis
Detection
Prevention or
Earliest Possible
Detection
Treatment
Monitor
Therapy
Change Treatment
Outcome
Acknowledgments
InterGenetics
University of California San Francisco
Sharmila Manjeshwar, PhD
Daniele DeFreese, MS
Bobby Gramling, MS
Thomas Pugh, MS
Laura Blaylock, BS
Kathie Dalessandri, MD
Margaret R. Wrensch, PhD
John K. Wiencke, PhD
Dr. Rei Miike, PhD
Statistical Consultants
Christopher C. Benz, MD
Christopher Aston, PhD
Dr. Lue Ping Zhou, PhD
Nicholas Knowlton, MS
OUHSC
John Mulvihill, MD
Buck Institute for Age Research
Zero Breast Cancer
Georgianna Farren, MD
Acknowledgments
Marin County Health Dept
Mark Powell, MD, MPH
Lee Ann Prebil, PhD
Rochelle Ereman, MS, MPH
Funding Sources
• NIH – SBIR
• US Army BCRP
• American Cancer Society
• Oklahoma Center for the Advancement of Science and Technology
• Swisher Family Trust
• Presbyterian Health Foundation
• Oklahoma Life Sciences Fund
Collaborators and Research Sites