Date held: February 12, 2015
Presented by: Deb Davison, Genomic Health
Topics discussed:
The latest in genomic testing and its role in cancer treatment
The most recent results from Genomic Health’s second independent clinical validation study of Oncotype DX® in DCIS patients
Q&A session about the implications of this research
2. Harnessing theHarnessing the
Power of GenomicsPower of Genomics
The OncoThe Oncotypetype DXDX®®
Breast Cancer Assay for DCISBreast Cancer Assay for DCIS
Deborah Davison, DNPDeborah Davison, DNP
Genomic Health Medical AffairsGenomic Health Medical Affairs
3. Genomic Health, Inc. – A Commitment to
Personalized Cancer Medicine
3
Do I needDo I need
ImmediateImmediate
therapy?therapy?
Do I needDo I need
chemotherapy?chemotherapy?
Do I needDo I need
radiation?radiation?
INVASIVEINVASIVE
BREAST CANCERBREAST CANCER
INVASIVEINVASIVE
BREAST CANCERBREAST CANCER
DCISDCIS
BREAST CANCERBREAST CANCER
DCISDCIS
BREAST CANCERBREAST CANCER PROSTATE CANCERPROSTATE CANCERPROSTATE CANCERPROSTATE CANCER
Do I haveDo I have
aggressiveaggressive
disease?disease?
STAGE II/IIISTAGE II/III
COLON CANCERCOLON CANCER
STAGE II/IIISTAGE II/III
COLON CANCERCOLON CANCER
2004 2011 2010 2013
Answers Key Critical Questions
4. The Oncotype DX®
assay is based on genomic science.
While genomics and genetics may sound similar and are
related, they focus on different information.
•GENETICS examines the function of a single gene
•GENOMICS examines groups of genes and their
relationships in order to identify their combined influence
4
Genetics vs. Genomics
5. For genomic assays, key questions…
• Does the assay provide insight into the
biology of the tumor?
• Does the assay provide information that we
didn’t already have?
• Is there clinical significance to the result?
• Can I be confident in the result?
6. Approximately 20% ofApproximately 20% of
All New Breast Cancers in the US Are DCISAll New Breast Cancers in the US Are DCIS
• Advances in technologies for screening and diagnosis have led to
an increase in detection of DCIS
– It is estimated that by 2020, over 1 million women in the US will
be living with a diagnosis of DCIS compared to 500,000 in 20051
BUT….
• Methods for assessing risk of local recurrence (LR) and making
treatment decisions have not kept pace with diagnostic advances
– Clinical and pathologic factors alone or combined by nomograms
rely on population-based estimates of average risk and have not
been able to separate which patients have a lower or higher risk
of recurrence
61.Allegra et al. J Natl Cancer Inst. 2010.
8. How Do You Make a Treatment Decision forHow Do You Make a Treatment Decision for
Your Patient with DCIS?Your Patient with DCIS?
1. Ernster et al. J Natl Cancer Inst. 2002. 2. NCCN Guidelines in Breast Cancer. v3.2014. 3. Fisher et al. J Clin Oncol. 1998. 4. Wapnir et al. J Natl Cancer Inst. 2011.
5. Bijker et al. J Clin Oncol. 2006. 6. Emdin et al. Acta Oncol. 2006. 7. McCormick et al. ASCO 2012.
• LR rates with surgery
alone range from 15-60% (about 50%
are invasive)2
Goals of DCIS therapy are variedGoals of DCIS therapy are varied
• Prevention of any LR and particularly an invasive LR is a primary considerationPrevention of any LR and particularly an invasive LR is a primary consideration
• Cosmetic outcomes: mastectomy vs. breast conserving surgery (BCS)Cosmetic outcomes: mastectomy vs. breast conserving surgery (BCS)
Multiple factors influence treatment decisionsMultiple factors influence treatment decisions
• Estimated risk of LR based on clinical and pathologic features
• Estimated risk of invasive LR (approximately 50% of recurrences)
• Balancing toxicity of therapy for a non-life threatening disease
• Patient preference
Treatments include:Treatments include:11
BreastBreast
conservingconserving
surgerysurgery
PartialPartial
or wholeor whole
breastbreast
radiationradiation
MastectomyMastectomy
PreventivePreventive
hormonalhormonal
therapytherapy
• Radiation therapy reduces LR by 50% but has not
been shown to impact overall or disease-free
survival3-7
8
9. The OncoThe Oncotypetype DXDX®®
Breast Cancer AssayBreast Cancer Assay
for DCIS:for DCIS:
A Genomic Risk Stratification ToolA Genomic Risk Stratification Tool
10. The Oncotype DX®
DCIS Assay
What is the test?
•A molecular diagnostic test that helps determine the
likelihood of DCIS returning or the tumor returning as invasive
breast cancer (local recurrence).
Who is the test for?
•Patients with DCIS (non-invasive) breast cancer (Stage 0)
that has not spread to the lymph nodes
Why do the test
•To determine if a patient is at a high risk of local recurrence
and whether those recurrences may be invasive
10
12. A Scientific Quote:
Sponge Bob Square Pants
Sandy the Squirrel scientifically describes how
a laugh is formed to Sponge Bob…
•Sponge Bob: “That sounds painful”
•Sandy: “Science makes everything sound painful”
13. Clinical Validation of theClinical Validation of the
OncoOncotypetype DXDX®®
Breast Cancer AssayBreast Cancer Assay
for DCIS:for DCIS:
The ECOG E5194 AnalysisThe ECOG E5194 Analysis
14. Summary: E5194 Validation of theSummary: E5194 Validation of the
OncoOncotypetype DXDX®®
Breast Cancer Assay for DCISBreast Cancer Assay for DCIS
• E5194 study validated the DCIS Score™
result as a strong
independent predictor of LR risk
– Any (DCIS or invasive) LR
– An invasive LR
• The DCIS Score result quantifies the 10-year risk of LR
– As a continuous variable or a categorical variable by 3
pre-specified risk groups
Solin et al. J Natl Cancer Inst. 2013. 14
15. E5194: Patient andE5194: Patient and
Tumor Characteristics of Analysis CohortTumor Characteristics of Analysis Cohort
Characteristic Number (N=327)
Patient age
Median
> 50 yr
61 yrs
261 (80%)
Postmenopausal 248 (76%)
Tumor size 7 mm (median)
Tumor size < 10 mm 260 (80%)
Negative margins > 5 mm 214 (65%)
Tamoxifen use 96 (29%)
ER positive (RT-PCR) 318 (97%)
Study cohort:
Cohort 1 (G1/2; ≤2.5 cm) 273 (83%)
Cohort 2 (G3; ≤1.0cm) 54 (17%)
Solin et al. J Natl Cancer Inst. 2013. 15
16. Solin et al. J Natl Cancer Inst. 2013.
E5194:E5194:
Pre-specified Study ObjectivesPre-specified Study Objectives
• Primary
– To determine whether there is a significant
association between the DCIS Score™
result and LR
risk
• Secondary
– To determine whether the DCIS Score result provides
value beyond standard clinical and pathologic factors
16
17. Solin et al. J Natl Cancer Inst. 2013.
The ECOG 5194 study validated the DCIS Score result as a predictor of any LR or
an invasive LR
•The DCIS Score result provides greater visibility into the risk of LR based on the
underlying tumor biology and separates patients with a lower risk from patients
with a higher risk of LR
DCIS ScoreDCIS Score™™
Result: 10-Year Local RecurrenceResult: 10-Year Local Recurrence
by Risk Group in E5194by Risk Group in E5194
17
Any Local Recurrence Invasive Local Recurrence
18. Solin et al. J Natl Cancer Inst. 2013.
The ECOG E5194 study validated the DCIS Score result as a predictor of
LR (increasing DCIS Score corresponds to increasing risk)
•Any DCIS or invasive LR
•An invasive LR
DCIS ScoreDCIS Score™™
Result: 10-YearResult: 10-Year
Local Recurrence in E5194Local Recurrence in E5194
18
Any Local Recurrence Invasive Local Recurrence
19. Hazard Ratio*
(95% CI)
P value
Primary (Univariable) Analysis
DCIS Score™
result 2.31 (1.15, 4.49) 0.02
Multivariable Analyses
Excluding the DCIS Score result
Tumor size 1.54 (1.14, 2.02) 0.006
Postmenopausal 0.49 (0.27, 0.90) 0.02
Including the DCIS Score result
DCIS Score result 2.37 (1.14, 4.76) 0.02
Tumor size 1.52 (1.11, 2.01) 0.01
Postmenopausal 0.49 (0.27, 0.90) 0.02
*Hazard ratio is for a 50-point difference. For study cohort, surgical margins, grade, comedo necrosis, and DCIS histologic pattern, all p > 0.46.
For tamoxifen use, p = 0.09. Since all were not significant, none of these factors were included in the multivariate analyses.
Solin et al. J Natl Cancer Inst. 2013.
The DCIS Score result is a strong independent predictor of LR risk
Risk for Local Recurrence in E5194: Primary andRisk for Local Recurrence in E5194: Primary and
Multivariable AnalysesMultivariable Analyses
19
20. Summary: E5194 Validation of theSummary: E5194 Validation of the
OncoOncotypetype DXDX®®
Breast Cancer Assay for DCISBreast Cancer Assay for DCIS
• E5194 study validated the DCIS Score™
result as a strong
independent predictor of LR risk
– Any (DCIS or invasive) LR
– An invasive LR
• The DCIS Score result quantifies the 10-year risk of LR
– As a continuous variable or a categorical variable by 3
pre-specified risk groups
Solin et al. J Natl Cancer Inst. 2013. 20
21. Second Validation of the OncoSecond Validation of the Oncotypetype DXDX®®
Breast Cancer Assay for DCIS:Breast Cancer Assay for DCIS:
The Ontario ProvincialThe Ontario Provincial
DCIS Cohort AnalysisDCIS Cohort Analysis
22. Clinical Implications: The DCIS ScoreClinical Implications: The DCIS Score™™
ResultResult
Provides Individualized Risk of Local RecurrenceProvides Individualized Risk of Local Recurrence
The second study reconfirms the association of the DCIS Score result
with LR as shown in E5194 and provides a strong, independent
predictor of LR risk
•Provides a quantitative and individualized estimate of LR risk that
goes beyond the risk associated with the traditional clinical and
pathologic factors
Oncotype DX®
helps patients and physicians better understand
their underlying tumor biology and can help guide treatment
decisions by distinguishing low-risk from high-risk DCIS
22Solin et al. J Natl Cancer Inst. 2013. Rakovitch et al. SABCS 2014.
23. An Observational Patient CohortAn Observational Patient Cohort
• Registry of 5,752 patients with DCIS collected from the Canadian
province of Ontario, between 1994 and 2003
– 3,795 women with DCIS had BCS with or without XRT (1,658
without XRT)
– All cases included in the primary analysis (BCS alone) had
central pathology review
• Analyses were prospectively designed as per standard methods
used in the first validation study (E5194)
Rakovitch et al. Breast Cancer Res Treat. 2013. 23
24. Ontario Cohort Study ObjectivesOntario Cohort Study Objectives
Primary Objective
•To evaluate if the DCIS Score™
result is associated with the risk of local
recurrence (DCIS or invasive) in patients treated with BCS alone with negative
margins (no tumor on ink) and no XRT
– Tested sequentially in ER-positive patients and then in all patients
regardless of ER status
Main Secondary Objectives
•To evaluate if the DCIS Score result is independently associated with local
recurrence adjusting for other clinical and pathologic factors
•To evaluate if the DCIS Score result is associated separately with the risk of
DCIS or invasive local recurrence
Rakovitch et al. SABCS 2014. 24
25. Characteristic N=571
Age Category Median 61 yrs
<50 yrs 110 (19.3%)
≥50 yrs 459 (80.7%)
Subtype Solid 358 (62.7%)
Cribriform 175 (30.6%)
Other 38 (6.7%)
Nuclear Grade Low 55 (9.6%)
Moderate 332 (58.1%)
High 184 (32.2%)
Rakovitch et al. SABCS 2014.
Ontario Cohort: Patient and TumorOntario Cohort: Patient and Tumor
CharacteristicsCharacteristics
25
Characteristic N=571
Comedo
Necrosis
Present 350 (61.3%)
Tumor Size
Category
≤10 mm 150 (26.3%)
>10 mm 140 (24.5%)
Missing 281 (49.2%)
Multifocality Present 114 (20.0%)
ER Status(PCR) Positive 541 (94.7%)
HER2 Status
(PCR)
Positive 100 (17.5%)
26. Primary Analysis: Association of the DCISPrimary Analysis: Association of the DCIS
ScoreScore™™
Result and Local Recurrence RiskResult and Local Recurrence Risk
Endpoint1
HR (95% C.I.)* P value*
Local recurrence in all
patients
2.15 (1.43, 3.22) <0.001
Local recurrence in ER+ DCIS 2.26 (1.41, 3.59) <0.001
1. Rakovitch et al. SABCS 2014. 2. Solin et al. J Natl Cancer Inst. 2013. 26
• The primary analysis showed that the DCIS Score result was a strong predictor of
LR in the group of patients that had BCS alone and negative margins, confirming
the results of E5194
• The association of the DCIS Score result with LR in the ER+ group was similar to
the association in the overall population, indicating that ER status was not a
driver of the score
*Cox model HRs for a 50-point difference in the DCIS Score result
• The HR for local recurrence in E5194 was 2.31 (1.15, 4.49)2
27. DCIS ScoreDCIS Score™™
Result: 10-Year Risk of Any LocalResult: 10-Year Risk of Any Local
Recurrence by Risk Group in the OntarioRecurrence by Risk Group in the Ontario
Provincial DCIS CohortProvincial DCIS Cohort
Rakovitch et al. SABCS 2014. 27
• The results confirmed the association of the DCIS Score result with LR and stratification
of recurrence risk based on underlying biology that is not apparent in the population as a
whole
• The proportion of patients within each risk group is also similar to what was observed in
the E5194 study with the majority of patients (62%) having a low score
DCIS Score Groups Continuous DCIS Score
28. DCIS ScoreDCIS Score™™
Result: 10-YearResult: 10-Year InvasiveInvasive oror DCISDCIS
Local Recurrence by Risk Group in the OntarioLocal Recurrence by Risk Group in the Ontario
Provincial DCIS CohortProvincial DCIS Cohort
Rakovitch et al. SABCS 2014. 28
• As in the E5194 study, this study showed that the DCIS Score result stratifies
patients for risk of an invasive LR
• Further, the DCIS Score result was able to stratify patients for risk of a DCIS LR
Invasive Local RecurrenceDCIS Local Recurrence
29. Multivariable Analysis: The DCIS ScoreMultivariable Analysis: The DCIS Score™™
ResultResult
Is an Independent Predictor of LocalIs an Independent Predictor of Local
RecurrenceRecurrence
Characteristic N
HR
(95% C.I.)
P value
DCIS Score /
50
571
1.68
(1.08, 2.62)
0.02
Age 0.03
≥50 459 1.0
<50 110
1.75
(1.07, 2.76)
Subtype 0.04
Cribriform 175 1.0
Solid 358
1.63
(0.97, 2.88)
Rakovitch et al. SABCS 2014. 29
DCIS Score result, tumor size, age, tumor subtype, and multifocality were all
independent predictors of LR risk
Characteristic N
HR
(95% C.I.)
P value
Tumor size 0.01
≤10mm 150 1.0
>10mm 140
2.07
(1.15, 3.83)
Multifocality 0.003
Absent/unkn
457 1.0
Present 114
1.97
(1.27, 3.02)
30. DCIS Score™
Result: Comparison of 10-Year Local
Recurrence:
ECOG E5194 and Ontario Cohort
Rakovitch et al. SABCS 2014.
The DCIS Score™
result stratifies patients consistently as shown in two
separate validation studies
E5194 Ontario Cohort
31. Clinical Implications: The DCIS ScoreClinical Implications: The DCIS Score™™
ResultResult
Provides Individualized Risk of Local RecurrenceProvides Individualized Risk of Local Recurrence
The second study confirms the association of the DCIS Score result
with LR as shown in E5194 and provides a strong, independent
predictor of LR risk
•Validated in 898 patients across two studies with consistent results
•Provides a quantitative and individualized estimate of LR risk that
goes beyond the risk associated with the traditional clinical and
pathologic factors
•Reflects the underlying tumor biology and can help guide treatment
decisions by distinguishing low-risk from high-risk DCIS
31Solin et al. J Natl Cancer Inst. 2013. Rakovitch et al. SABCS 2014.
32. Consistent Risk Stratification by the DCIS ScoreConsistent Risk Stratification by the DCIS Score™™
Result in Patient Cohorts from Validation StudiesResult in Patient Cohorts from Validation Studies
and Genomic Health Laboratoryand Genomic Health Laboratory
67.4%
15.3%
17.3%
13.5%
16.2%
70.3%
62.2%
21.2%
16.6%
1. Sing et al. ESMO 2014. 2. Solin et al. J Natl Cancer Inst. 2013 3.Rakovitch et al. SABCS 2014. 32
33. The DCIS ScoreThe DCIS Score™™
Result ProvidesResult Provides
Information Beyond Clinical andInformation Beyond Clinical and
Pathologic FactorsPathologic Factors
Comparing risk estimates of the DCIS Score result toComparing risk estimates of the DCIS Score result to
grade, age, size, DCIS pattern, and comedo necrosisgrade, age, size, DCIS pattern, and comedo necrosis
34. 34
CAP Nuclear Grade CAP Nuclear Grade
DCIS Score
There Is a Broad Range of DCIS ScoreThere Is a Broad Range of DCIS Score™™
ResultsResults
Across Tumor Grade in E5194Across Tumor Grade in E5194
• There was no association between nuclear grade and LR risk
• There was a broad range of DCIS Score results across each grade level
Solin et al. J Natl Cancer Inst. 2013.
35. There is a Broad Range of DCIS ScoreThere is a Broad Range of DCIS Score™™
ResultsResults
Across Clinical and Pathologic Characteristics inAcross Clinical and Pathologic Characteristics in
E5194E5194
Solin et al. J Natl Cancer Inst. 2013. 35
Comedo Necrosis
Menopausal StatusDCIS Pattern
Tumor Size
36. Rakovitch et al. Data on file.
36
There is a Broad Range of DCIS ScoreThere is a Broad Range of DCIS Score™™
ResultsResults
Across Clinical and Pathologic Characteristics inAcross Clinical and Pathologic Characteristics in
the Ontario Provincial Cohortthe Ontario Provincial Cohort
Tumor Size
DCIS Pattern
Comedo Necrosis
Absent
Present
Age
37. The DCIS ScoreThe DCIS Score™™
Result Provides InformationResult Provides Information
Beyond Clinical and Pathologic FactorsBeyond Clinical and Pathologic Factors
37
• Across two validation studies with different patient populations,
there was a consistent and broad range of DCIS Score results
across clinical and pathologic features
Age, grade, tumor size, comedo necrosis, and DCIS
pattern alone cannot predict the DCIS Score result
Solin et al. J Natl Cancer Inst. 2013. Rakovitch et al. SABCS 2014.
Because each clinical or pathologic factor has a range of DCIS
Score results, the factors cannot be used as a surrogate or
predictor of the DCIS Score result
38. Impact of the DCIS ScoreImpact of the DCIS Score™™
Result onResult on
Physicians’ TreatmentPhysicians’ Treatment
RecommendationsRecommendations
A U.S. Multicenter StudyA U.S. Multicenter Study
39. Study DesignStudy Design
Primary objective: to estimate the proportion of patients for whom the
DCIS Score™
result led to a change in the recommendation for XRT
• Patient meeting
study inclusion
criteria is enrolled
•Patient characteristics
•Pathology; ERPR
•Treatment
recommendation
• DCIS Score report results
• Treatment recommendation
• Factors affecting physician
recommendations
DCIS Score report
• 10 US centers; 115 patients included in the analysis
• Representative of a contemporary patient population
• Majority of patients – post-menopausal, ER+, < 2cm
• Both physician groups represented: 5 Rad Oncs, 5 Surg Oncs
Patient characteristics
Alvarado et al. ASCO 2014.
Prospective
Enrollment
Collected
Pre-Assay
Data Collected
Post-Assay
Data
39
40. The DCIS ScoreThe DCIS Score™™
Result Impacts RecommendationResult Impacts Recommendation
for XRT by Revealing Underlying Biologyfor XRT by Revealing Underlying Biology
84 (73.0%)
68 (59.1%)
31 (27.0%) 47 (40.9%)
Alvarado et al. ASCO 2014. 40
• The DCIS Score result changed the recommendation for XRT 31% of the time
(p= 0.008; McNemar’s test)
• This degree of change reflects the impact of the additional information
regarding the risk of LR and the individual underlying tumor biology that is
not evident with the clinical and pathologic features
41. Changes in Recommendation for XRT withinChanges in Recommendation for XRT within
DCIS ScoreDCIS Score™™
GroupsGroups
Alvarado et al. ASCO 2014.
Numberofpatients
XRT Recommendation
Pre-
Assay
Pre-
Assay
Pre-
Assay
Post-
Assay
Post-
Assay
Post-
Assay
Low Intermediate High
DCIS Score Group
41
42. The OncoThe Oncotypetype DXDX®®
Breast Cancer AssayBreast Cancer Assay
for DCIS Report: A Tool for Sharedfor DCIS Report: A Tool for Shared
Treatment DecisionsTreatment Decisions
42
43. Clinical Utility of the OncoClinical Utility of the Oncotypetype DXDX®®
Breast Cancer Assay for DCISBreast Cancer Assay for DCIS
A Paired Case StudyA Paired Case Study
44. PATIENT A
55-year-old patient with 1.4-cm tumor
Menopausal Status: Postmenopausal
Tumor Type: DCIS
Tumor Size: 1.4 cm
ER Status (IHC): Positive
PR Status (IHC): Positive
Histologic Grade: 2
General Health: Good
PATIENT B
60-year-old patient with 1.6-cm tumor
Menopausal Status: Postmenopausal
Tumor Type: DCIS
Tumor Size: 1.6 cm
ER Status (IHC): 90% Positive
PR Status (IHC): 98% Positive
Histologic Grade: 3
General Health: Hypertension, diabetes,
morbid obesity
Cases are based on actual patients and are used for illustrative purposes.
Paired Case PresentationPaired Case Presentation
44
45. PATIENT A RESULTS
Clinical Experience
Patients with a DCIS Score of 56 had a 22% risk of
any local recurrence and a 12% risk of an invasive
local recurrence.
PATIENT B RESULTS
Clinical Experience
Patients with a DCIS Score of 24 had a 13% risk of
any local recurrence and a 6% risk of an invasive
local recurrence.
Paired Case PresentationPaired Case Presentation
45
46. Harnessing the Power ofHarnessing the Power of
Genomics for Personalized Management of DCISGenomics for Personalized Management of DCIS
The Oncotype DX®
Breast Cancer Assay for DCIS is
an important advancement in providing an individualized risk of local
recurrence and personalizing treatment for patients with DCIS
46
48. Canadian Cancer Survivor Network
Contact Info
Canadian Cancer Survivor Network
1750 Courtwood Crescent, Suite 210
Ottawa, ON K2C 2B5
Telephone / Téléphone : 613-898-1871
E-mail jmanthorne@survivornet.ca or mforrest@survivornet.ca
Web site www.survivornet.ca
Blog: http://jackiemanthornescancerblog.blogspot.com/
Twitter: @survivornetca
Facebook: www.facebook.com/CanadianSurvivorNet
Pinterest: http://pinterest.com/survivornetwork/
Editor's Notes
GHI10429_1114
Advances in diagnosis and screening have led to an increase in the number of patients with DCIS. In 2005, 500,000 women were estimated to be living in the US with a diagnosis of DCIS; this number is expected to reach 1 million in 2020.
However, treatment decision making has not evolved to keep pace with the advances in detection of these tumors. The traditional measures for assessing risk of recurrence for DCIS are similar to the clinical and pathologic measures used to assess risk of recurrence in invasive breast cancer: age, residual tumor/margin width, grade, histology, size and menopausal status. None of these characteristics, however, provide a quantitative assessment of recurrence risk.
Key Points: The DCIS algorithm was developed based upon an analysis of multiple correlative science studies and evidence suggesting that there is a biologic relationship between DCIS and invasive carcinoma.
Initially, the Oncotype DX assay was performed and the expression levels of the 21 genes were compared for manually micro-dissected DCIS and invasive carcinoma when both were present within the same fixed paraffin-embedded tumor tissue (FPET). A wide range of single gene results and Recurrence Scores were noted in the DCIS components – all DCIS did not have low Recurrence Scores. A strong correlation between gene expression profiles was found for paired invasive and DCIS components; however, the proliferation group scores were lower for the DCIS components.
Selection of genes and algorithm development for the DCIS Score was based on published results for the Oncotype DX assay for invasive breast cancer, examining correlations with both local recurrence and distant recurrence. Seven cancer-related genes (the five proliferation genes, PR, and GSTM1) and the five reference genes were chosen for their ability to predict recurrence, regardless of adjuvant tamoxifen use. The proliferation group score uses the continuous proliferation score which provides important information for predicting IBE risk. The 12 selected genes are a subset of the 21 genes included in the Oncotype DX Recurrence Score.
After the algorithm was developed, a cohort of pure DCIS samples was analyzed and a similar wide range of gene expression was observed (Genomic Health, Inc. internal data on file). This cohort was used to scale the DCIS Score to range from 0 to 100, with a higher score indicating a greater likelihood of recurrence.
This presentation is focused upon harnessing the power of genomics for DCIS management. Foremost in the clinician’s mind is the question of how to make a treatment decision for the patient. These decisions vary according to patient preference, and goals of treatment must be aligned with the patient’s own goals.
The decision around recommending XRT is dependent on an assessment of LR risk with an assumption that around half of those recurrences will be invasive disease. While reduction of LR is important, particularly invasive LR, other goals of therapy are taken into consideration such as the cosmetic outcomes and the side effects from XRT. Currently LR risk is estimated based on clinicopathologic factors and provides an average risk derived from population studies.
No studies have been able to identify a patient group that did not derive benefit (i.e. risk reduction) with XRT.
Now we will discuss the only genomic assay that can assess local recurrence risk in DCIS, whether it is invasive or DCIS.
The DCIS Score algorithm includes 5 genes in the proliferation group, PR, GSTM1, and 5 reference genes.
Similar to the Recurrence Score® result, the DCIS Score result is scaled as a continuous variable from 0-100.
The pre-specified DCIS Score risk groups are:
Low &lt; 39, Intermediate 39 – 54, High ≥ 55
The E5194 validation study showed that the DCIS Score result quantifies the 10-year risk of LR (any LR or invasive LR) as a continuous variable or a categorical variable by 3 pre-specified risk groups.
It has been argued that the E5194 patient population is not fully representative of modern clinical practice because it was believed to be at perceived low risk for local recurrence due to small tumor size, large margin width and low grade.
The parent E5194 trial enrolled patients with either low- or intermediate-grade DCIS measuring 2.5 cm or smaller, or high-grade DCIS measuring 1 cm or smaller who had microscopic margin widths of 3 mm or wider and no residual calcifications on postoperative mammograms. Patients entered in 2000 and later could take tamoxifen if they wished (Hughes LL et al. J Clin Oncol. 2009; 5319-5324).
The patient population in the validation study cohort was similar to the parent trial:
The average tumor size in the validation study was slightly larger than in the parent trial. The median size in the parent trial was 6 mm for low/intermediate grade (cohort 1) and 5 mm for the high grade (cohort 2).
A total of 29% of patients received tamoxifen.
97% of patients were ER-positive by RT-PCR.
The primary analysis was pre-specified as part of the statistical analysis plan, which was finalized prior to the merging of clinical and Genomic Health’s laboratory data.
The primary study objective was to determine whether there is a significant association between the DCIS Score result and the risk of LR (DCIS or invasive carcinoma).
The E5194 clinical validation study was a prospectively defined analysis using archived tissues from a previously completed clinical trial (parent E5194). Thus, the design of the study is called a prospective-retrospective clinical validation study.
The primary objective was to demonstrate that the DCIS Score result was significantly associated with the risk of LR (DCIS or invasive), analyzed as both a continuous variable and a categorical variable based on the 3 risk groups (low &lt;39, Intermediate 39-54 and high ≥55).
The secondary objective was to determine whether the DCIS Score result provided value beyond standard clinical and pathologic factors.
This slide shows the Kaplan-Meier estimates of the risk of any local recurrence (left panel) and the risk of invasive local recurrence (right panel) over time for each of the three DCIS Score groups.
In the high risk group (red), the estimated 10-year percentage of patients free of any LR was 74%, with the low DCIS Score group (green) having an estimated 89% of patients free of LR. The log rank p-value of 0.006 means that there was a significant trend in LR risk across the three DCIS Score groups.
For invasive local recurrence, the high DCIS Score group (red) had an estimated 81% of patients LR-free at 10 years, while in the low DCIS Score group (green), 96% of patients were estimated to be LR-free at 10 years. The log rank p-value of 0.003 means that there was a significant trend in invasive LR risk across the three DCIS Score groups.
While the E5194 population was generally believed to be at low risk, it is important to note that 70% of the patients were in the low DCIS Score group and 30% were in the intermediate or high risk groups.
The estimation of 10-year risk of of local recurrence risk evaluated as a continuous function shows that increasing DCIS Score is associated with increasing risk of local recurrence (any recurrence or invasive recurrence) in the E5194 study using a Cox proportional hazards model.
The validation study met its primary objective. The DCIS Score result predicted the likelihood of any local recurrence with a significant p-value of 0.02 and HR of 2.31.
Multivariable analysis showed that the DCIS Score result was a strong predictor of local recurrence risk.
In multivariable analysis, excluding the DCIS Score, only tumor size and post-menopausal status were significant predictors for the risk of local recurrence. The traditional clinical and pathologic variables, including surgical margins, nuclear grade, comedo necrosis, DCIS pattern (papillary, micropapillary, cribiform and solid) and tamoxifen use, were not statistically significant in this analysis.
The DCIS Score result was added to this multivariable model, with the results shown here (in the model “including the DCIS Score”). This shows that the DCIS Score result was able to predict LR risk over and above the traditional factors, tumor size and menopausal status, with an estimated hazard ratio of 2.37 associated with a 50 point difference in the DCIS Score result. This hazard ratio is unchanged from the model that does not include any additional factors, showing that the DCIS Score result is essentially independent of these factors as a predictor of LR risk.
The E5194 validation study showed that the DCIS Score result quantifies the 10-year risk of LR (any LR or invasive LR) as a continuous variable or a categorical variable by 3 pre-specified risk groups.
The second study confirmed the association of the DCIS Score result with local recurrence as shown in E5194 in this broader patient population.
The second validation was carried out in a real-world patient cohort that was not restricted as to margins, tumor grade, or size. Therefore, this is a more stringent test of the performance of the assay.
As with the first validation study, the primary analysis was pre-specified as part of the statistical analysis plan, which was finalized prior to the merging of clinical and Genomic Health’s laboratory data.
The second validation study was carried out in a broad DCIS patient population representative of community practice in Ontario.
The primary analyses in both validation studies were similar and confirmed that the assay is able to stratify patients for risk of local recurrence.
The association of the DCIS Score result with LR in the ER+ group was similar to the association in the overall population suggesting that the DCIS Score result provides information that is independent of ER expression.
For comparison, in E5194, 97% of patients were ER+.
The DCIS Score result was associated with LR and was able to risk stratify patients based on underlying biology that was not apparent in the population as a whole (white line; all patients).
As in the E5194 study, the Ontario study showed that the DCIS Score result could stratify patients for invasive local recurrence. Further, shown in this study, and in contrast to E5194, the assay can also risk stratify patients for a DCIS local recurrence.
DCIS Score result, multifocality, tumor size, age, and tumor subtype were all independent predictors of LR risk.
As in the E5194 study, the Ontario study showed that the DCIS Score result could stratify patients for invasive local recurrence. Further, shown in this study, and in contrast to E5194, the assay can also risk stratify patients for a DCIS local recurrence.
The second study confirmed the association of the DCIS Score result with local recurrence as shown in E5194 in this broader patient population.
In the first two years of availability, almost 4,000 samples were sent for a DCIS Score result. The distribution of scores in this group is similar to the validation studies, with the majority of patients at low risk for local recurrence.
As shown in the two panels on this slide, there is no association between grade and the risk of LR. There is a distribution of DCIS Score results across grade in the E5194 population.
A wide range of DCIS Score results was observed within each category.
Size: Tumor size as a categorical variable was not a significant predictor of local recurrence risk, however, it was as a continuous variable.
Tumor size analyzed as a categorical variable was not statistically significant, with p=0.28 in the log rank test. When tumor size was analyzed as a continuous variable, it was a significant predictor of LR risk in univariable analyses and maintained significance in multivariable analyses.
There was a poor correlation between the DCIS Score result and tumor size, with a Spearman’s rank correlation coefficient of 0.18.
Accurate measurement of tumor size can be challenging and may vary widely from pathologist to pathologist. Thus, the DCIS Score result which provides a reproducible assessment of the risk of LR may be useful to inform treatment recommendations.
Menopausal Status: In addition to grade, age and menopausal status have been used to assess overall risk of LR for patients with DCIS. Menopausal status is a significant factor in predicting the risk of LR; however, there was no correlation between menopausal status and the DCIS Score result.
Comedo Necrosis: There was a moderate association between the DCIS Score result and comedo necrosis with a Spearman’s rank correlation coefficient of 0.49.
Pattern:
There was no association between the DCIS Score result and DCIS pattern.
As in the E5194 population, there was a distribution of DCIS Score results across clinical and pathologic categories in the Ontario second validation study.
Across two validation studies with different patient populations, there was a consistent and broad range of DCIS Score results across clinical and pathologic features.
This was a prospectively enrolled observational study of newly diagnosed patients with histologically proven pure DCIS; 115 patients were evaluated.
122 patients were enrolled at 10 centers throughout the US from September 2012 to February 2014
115 patients were evaluable for the primary analysis
The 122 patients were enrolled by 5 radiation oncologists (48 pts; 41.7%) and 5 surgeons (67 pts; 58.3%)
Sites: RMCC, Denver CO- 48 pts; SCCA, Little Rock, AR- 28 pts; CH, Cincinnati, OH- 15 pts; UCSF, San Francisco, CA- 9 pts; SJMC, Baltimore, MD- 9 pts; SEH, Cincinnati, OH- 6 pts
Physicians filled out standardized questionnaires prior to and after the DCIS Score results were known; patient and tumor characteristics were extracted from the medical record
Inclusion Criteria:
≥18 years old, female
Histologically proven DCIS
Eligible for breast conserving therapy
Surgical excision pathology report available
Oncotype DX® Breast Cancer Assay for DCIS ordered but result not yet available
Exclusion Criteria:
LCIS without DCIS
Invasive carcinoma
Mastectomy planned
Prior to receiving the DCIS Score result, physicians recommended XRT for 73% of patients. After receiving the assay results, XRT was recommended only for 59.1% of patients. This resulted in a net change of 31.3% in treatment recommendations.
The change in XRT recommendation by DCIS Score result had the greatest impact in the low risk group; overall, treatment recommendations changed 31.3% of the time.
The report is a tool that can be used in patient discussions to facilitate shared treatment decisions.
The DCIS Score™ result facilitates treatment decisions by increasing clarity, comfort and confidence in your patient’s treatment plan.