View
133
Download
2
Tags:
Embed Size (px)
DESCRIPTION
Citation preview
Centre de Lutte contre le Cancer d'Auvergne Clermont-Ferrand - France -
Centre Jean Perrin
Un test décentralisé apporte t il une valeur ajoutée aux équipes d’oncologie médicale dans les cancers du sein en
situa8on adjuvante ? Reproduc8bilité et fiabilité des résultats
Frédérique Penault-Llorca, MD, PhD
CENTRALIZED APPROACH
The Oncotype DX® Assay
Genomic Health, Inc.
OncotypeDX (Genomic Health, USA)
HR+ / HER2- , T1-3, N-/N+ FFPE specimens
qRT-PCR 21 GENES
PROLIFERATION, OESTROGENE, HER2, INVASION (16 GENES) + REFS (5 GENES)
« CENTRALIZED » TEST
(recurrence score) RS Late recurrence (10 years)
Benefit from adjuvant TT PROGNOSTIC AND PREDICTIVE
LOW RISK :
+ HORMONOTHERAPY / - CHEMOTHERAPY
INTERMEDIATE RISK : DISCUSSION
HIGH RISK : + HORMONOTHERAPY / + CHEMOTHERAPY
5
The Oncotype DX® assay is analy3cally validated
Elements of analy8c valida8on • Analy3cal sensi3vity (limits of detec3on and quan3ta3on)
• Assay precision and linear dynamic range • Analy3cal reproducibility • PCR amplifica3on efficiency • Sample and reagent stability • Reagent calibra3on • Instrument valida3on and calibra3on
Chau CH, et al. Clin Cancer Res. 2008;14(19):5967-5976.
Analytical validation is the assessment of assay performance characteristics and the optimal conditions to
generate accuracy, precision and reproducibility
MammaPrint®
Agendia, Inc.
MammaPrint (Agendia, NL)
HR+ ET HR - / HER2- , T < 5cm, N ≤ 3
Fresh frozen=> FFPE DNA array
70 GENES CELL CYCLE/ PROLIFERATION
SIGNAL TRANSDUCTION INVASION, METASTASIS, ANGIOGENESIS
« CENTRALIZED » TEST
RECENTLY ADAPTATED TO FFPE
Group of genes (« signatures »)
EARLY RECURRENCE (Dg < 5 ans) PROGNOSTIC
GOOD SIGNATURE : LOW RISK
POOR SIGNATURE : HIGH RISK
HR+& HR-‐
70 Gene Assay FFPE Uncertainty • FFPE tumor 3ssue fixa3on
causes RNA to degrade, the accuracy of microarray tes3ng depends on keeping the tumor RNA intact
• The 70 gene assay analy3cal validity tests were performed on fresh frozen 3ssue (current method in the FDA label)
• The 70 gene assay is now available in paraffin as part of the SYMPHONY tests; however, adequate valida3on of this method is not documented in the public literature
h\p://www.agendia.com Scicchitano MS, et al. J. Histochem. Cytochem. 2006; 54 (11): 1229–1237. 8
Procedure
• FFPE • Fill the form • Send block, slides or tumor
sample to the central lab with a dedicated box
• Results within 8 days via e-‐mail
DECENTRAL GENE EXPRESSION ANALYSIS
Prosigna PAM50 ROR NanoString nCounter®
12
Development of Prosigna™ is Based on PAM50 Gene Signature
2000 Researchers first describe
breast cancer intrinsic subtypes based on microarray
experiments
2009 Researchers first describe “PAM50” gene expression
signature
2010 NanoString exclusively licenses
PAM50 gene expression signature
2012/13 Prosigna launches after receiving CE Mark for Europe & Israel; FDA 510k clearance in US
PAM50 developed by a consor3um of four academic breast cancer experts ● Charles Perou, PhD, University of North Carolina
● Dr. Ma\ Ellis, Washington University School of Medicine
● Torsten Nielsen, MD, PhD, Pathologist, BC Cancer Agency
● Philip Bernard, MD, University of Utah / Huntsman Cancer Ins3tute
Source: Molecular portraits of breast cancer. Nature. 2000 May 25;. Source: Supervised Risk Predictor of Breast Cancer Based on Intrinsic Subtypes, JCO.2009
Overview of Principles: Design Advantages of nCOUNTER • Direct detec8on (no amplifica3on of target)
• Designed for short sequences ~100 bp
• Digital coun8ng results in excellent analy8cal performance
– Highly sensi3ve and precise – Wide dynamic range (5 logs)
• Automated processing
• Internal controls • Mul8plexed
Capture Probe
Reporter Probe
Target
Target-Probe Complex
SOLUTION HYBRIDIZATION
REMOVE EXCESS PROBE IMMOBILIZE/ALIGN
DIGITAL COUNT
13
14
Intrinsic Subtype: Organizing Framework for Breast Cancer
Supported by The Cancer Genome Atlas Study1
● Endocrine therapy alone Luminal A
1. Comprehensive molecular portraits of breast cancer. Nature. 2012 Oct 4;490(7418):61-‐70. 2. Personalizing the treatment of women with early breast cancer: highlights of the St Gallen InternaQonal Expert Consensus on the Primary
Therapy of Early Breast Cancer 2013 Annals of Oncology Advance Access published August 4, 2013
Diverse genetic and epigenetic alterations converge phenotypically into the four main breast
cancer subtypes defined by PAM50
Endorsed in 2013 St. Gallen Guidelines2
● If HER2―, endocrine +/-‐ cytotoxic therapy ● If HER2+, cytotoxics + an3-‐HER2 +
endocrine ● Could include anthracyclines and taxanes
Luminal B
● Cytotoxics + an3-‐HER2 ● Could include anthracyclines and taxanes
HER2 enriched
● Cytotoxics therapy alone, poten3ally including anthracyclines, taxanes and analkyla3ng agent
● Do not rou3nely use cispla3n or carbopla3n
Basal-‐like
15
Three Elements of Prosigna™ Breast Cancer Assay
Hardware: nCounter Analysis System
Consumable: Prosigna Kits
SoYware: Prosigna Report
Prep Sta8on
Digital Analyzer
Includes: ● 50 gene-‐based CodeSet with 8 controls ● Other consumables required for assay ● CE Marked Roche RNA isola3on kit sold separately
nCounter Analysis System and Prosigna Breast Cancer Assay Kit received FDA 510K clearance in 2013 and CE Marked in 2012
16
Prosigna™ Tests Formalin-Fixed Paraffin-Embedded Samples
H&E stain to iden3fy tumor area and
cellularity
Tumor area transposed to unstained slides and macrodissected
RNA extracted Block selected
Specimen Attribute Requirement Tissue input Viable invasive breast carcinoma (ductal, lobular, mixed, or NOS/NST)
Tissue input format Macrodissected 10-micron-thick slide-mounted tissue sections
Minimum tumor size 4 mm2 tumor area
Minimum tumor cellularity 10% within tumor area
Minimum RNA amount 125 ng (12.5 ng/µl)
Tissue area ≥100mm2 1 slides required
20 – 99mm2 3 slides required
4 – 19mm2 6 slides required
17
Simple and fast workflow is well suited for qualified clinical laboratories
Simple Prosigna™ Workflow Enables Decentralized Tes3ng Model
1
nCounter® Prep Station nCounter® Digital Analyzer
Hybridize 2 Purify 3 Count
Step 3 3 – 4.5 HOURS, AUTOMATED
5 min
HANDS-ON Step 2 2.5 – 3.0 HOURS, AUTOMATED
5 min
HANDS-ON Step 1 12 HOURS OR OVERNIGHT
5 min
HANDS-ON
PAM50 ROR by NanoString nCounter®
Extract RNA from FFPE
tumor sample
Run RNA & PAM50 CodeSet on nCounter Analysis System
Capture paQent expression profile
Calculate Risk of Recurrence (ROR) Score
Determine Intrinsic Subtype through Pearson’s CorrelaQon to Centroids
4 or 10 samples Overnight incubation
19
PAM50 Algorithm Generates a Prosigna Score for Each Patient
● Gene expression data are weighted with clinical variables to determine an integer score from 0 through 100 (ROR/Prosigna Score) indica3ve of the probability of distant recurrence
● ROR is based on the similarity of the gene expression profile to intrinsic subtypes, prolifera3on score, and tumor size
● Assay requires input of gross tumor size and nodal status Determine intrinsic subtype through Pearson’s correla8on to centroids
ROR = aRLumA+
bRLumB+
cRHer2e+
dRBasal+
eP+
fT
Pearson’s correla3on to centroids
Calcula8ng ROR (Prosigna Score)
Pa8ent expression profile
Prosigna centroids
Prolifera3on score Gross tumor size >2cm
Gnant M, et al. SABCS 2012; poster P2-10-02.
20
Patient Report Output: Page 1
Patient report: Identifying information
Assay description: Describes components of the Prosigna assay
Risk of Recurrence: Patient specific ROR is reports based on Prosigna algorithm. The ROR ranges from 0 to 100
Probability of distant recurrence: This section provides the correlation of the ROR with a specific likelihood of distant recurrence at 10 years, based on the average 10-year distant recurrence rate for that ROR in the clinical trial population. The probability of distant recurrence at 10 years increases continuously with an increase in ROR
Designed as a tool for pa8ent/oncologist communica8on
CE-‐IVD-‐marked
21
Patient Report Output: Page 2
Description of validation studies
Distant recurrence by subtype
Risk curves by study
Provided as detailed background on valida8on studies for oncologist
CE-‐IVD-‐marked
22
Highlights of Prosigna™ Report
Prosigna™ Output: Risk Interpreta3on and Categoriza3on by Nodal Status
• Risk classifica3on guidelines are provided based on cutoffs related to clinical outcome in the tested pa3ent popula3ons:
• 10-‐year probability of distant recurrence of < 10% is considered low risk • 10-‐year probability of distant recurrence of > 20% is considered high risk • Prosigna score discriminates risk groups within pa3ents with
node-‐nega3ve disease
23
Nodal Status Prosigna Score Range
Risk Categoriza8on
Node-‐nega3ve
0 -‐ 40 Low
41 -‐ 60 Intermediate
61 -‐ 100 High
Node-‐posi3ve (1 -‐ 3 nodes)
0 -‐ 40 Low
41 -‐ 100 High
Prosigna Package Insert.
24
Risk Categories Map to Clinical Risk by Nodal Status (0-‐10%, 10-‐20%, >20% 10 year DR)
10-Y
ear P
roba
bility
of D
istan
t Rec
urre
nce
ROR Score
0 10 20 30 40 50 60 70 80 90
100
0 20 40 60 80 100
Rate 95% CI
Node Negative
10-Y
ear P
roba
bility
of D
istan
t Rec
urre
nce
0 10 20 30 40 50 60 70 80 90
100
0 10 20 30 40 50 60 70 80 90 100
Rate
ROR Score
Node Positive (1-3 nodes) Low Int High 0-‐10 10-‐20 <20
Low Int High 0-‐10 10-‐20 <20
Nodal status
ROR range
Risk categoriza8on
Node-‐nega3ve
0-‐40 Low
41-‐60 Intermediate
61-‐100 High
Node-‐posi3ve (1-‐3 nodes)
0-‐15 Low
16-‐40 Intermediate
41-‐100 High
Prosigna™ Analytical Validation : Reproducibility and Precision Evaluated in Two Studies :
25
Study 1: Reproducibility from 8ssue
Extract RNA from FFPE tumor sample
Study 2: Precision from RNA
Prosigna Score
Run Prosigna on nCounter Dx Analysis
System
26
Analytical validation
27
Prosigna™ Analytically Validated for Decentralized Testing
Reproducibility from 43 FFPE Tissue Samples, Replicates across 3 different sites & n-‐counters 1
Precision from 108 replicates of 5 Pooled RNA samples, at 3 differents sites, 2 operators at each site, each with 3 differents reagants lots, 9 runs per operator 1
● For intrinsic subtype classifica8ons, the average concordance between sites was 97%
● Prosigna Score Standard Devia8on = 2.9 (scale 0-‐100)
● 100% concordance between the subtype & risk groups
● Site-‐to-‐site or operator-‐to-‐operator <1% of variance
● Prosigna Score Standard Devia8on = 0.67 (scale 0-‐100) 1 AnalyQcal Reproducibility of the Breast Cancer Intrinsic Subtyping Test and nCounter® Analysis System Using Formalin-‐Fixed Paraffin-‐Embedded (FFPE) Breast
Tumor Specimens. T Nielsen et al., S McDonald, S Kulkarni, J Storhoff, C Schaper, B Wallden, S Ferree, S Liu, V Hucthagowder, K Deschryver, V Holtschlag, G Barry, M Evenson, N Dowidar, M Maysuria, D Gao USCAP 2013
Assay Robust Against Non-tumor Tissue
RO
R C
hang
e
Percent Non-tumor
" Objective : o Assess impact of adjacent non-tumor tissue
on ROR. " Design:
o Slide mounted sections from 24 FFPE blocks were tested with vs. without macrodissection of adjacent non-tumor tissue.
o The difference in ROR between the macrodissected vs. unmacrodissected tissue was determined.
" Result: o The NanoString test result was robust
against the inclusion of up to 50% adjacent non-tumor tissue into the assay.
1 AnalyQcal Reproducibility of the Breast Cancer Intrinsic Subtyping Test and nCounter® Analysis System Using Formalin-‐Fixed Paraffin-‐Embedded (FFPE) Breast Tumor Specimens. T Nielsen et al., USCAP 2013
ENDOPREDICT (MYRIAD GENETICS)
EndoPredict (Myriad genetics)
HR+ / HER2- , T1-2, N0
FFPE qRT-PCR
7 GENES SIGNATURE PROLIFERATION, OESTROGENES
« LOCAL » TEST (SPECIAL EQUIPMENT IS REQUIRED)
SCORE OF RECURRENCE EP SCORE
LATE AND EARLY RECURRENCES (5 & 10 YEARS) PROGNOSIS
LOW RISK
HIGH RISK
UBE2C BIRC5 DHCR7
STC2 AZGP1 IL65T
RBBP8 MGP
Analytical validation according to CLSI guidelines is published in peer-reviewed journal
Peer-‐reviewed publica8on CE-‐IVD-‐marked
IVDD 98/79/EG
Analytical Performance Characteristics Core biopsies and surgical specimen can be used for EndoPredict
• Comparable results between core biopsies and surgical sec3ons • Inflammatory changes induced by presurgical biopsies had no significant
effect on the EndoPredict-‐based risk assessment in surgical specimens
CONCLUSION
Central versus local • Pros
– Central • Standardiza3on • High volume • High turnaround 3me • Easy shipping
– Local • Independency from large
companies • Ins3tu3onal based result • Cross lab valida3on • Fits into the INCa’s plaxorms
model • The automate is flex can be
used for research
• Cons – Central
• Usually Abroad • Absence of cross lab valida3on
– Local • Quality assurance • Turnaround 3me vs cost
effec3veness (30 samples/round)
• Implementa3on in a path lab rou3ne