Diagnostic Medicine

1
Diagnostic Medicine Digital Pathology Genomic
Classifiers Individualized Breast Cancer
Treatment Decisions

• Rick Baehner, MD
• Assistant Clinical Professor UCSFDirector of
  IHC Core Laboratory
• UCSF/Mt Zion Cancer Center
• Chief of Pathology
• Genomic Health
• Pathology Visions 2008

2
Outline

• Breast Pathology - yesterday, today tomorrow
• Highlight how Digital Pathology would be useful
• Discussion of quantitative prognostic and
  predictive molecular tools in breast cancer
• Digital Pathology
• Current uses of digital pathology at Genomic
  Health
• Clinical Laboratory
• Clinical Studies/Biomaterial Repository (BMR)
• Planned future applications of digital pathology
• Clinical Laboratory
• Clinical Studies

3
p???? Pathos

• p???? fate or harm
• Cancer the crab
• Anatomic Pathology Study of Disease provided
  the foundation of modern medicine
• First postmortem dissections by the Arabian
  physician Avenzoar (10911161)
• Rudolf Virchow (1821-1902) is recognized father
  of microscopic pathology

Most early pathologists were practicing
physicians or surgeons
4
Pathology The Past
Robert Hooke's microscope

• Invention of Microscope in 16th Century
• Discovery of cells, proteins, DNA mitotic
  figures the development of histochemical
  stains, e.g., eosin hematoxylin

Old Technologies lenses from 1600s
http//en.wikipedia.org/wiki/Microscopes
5
Pathology The Past

• Schism Anatomists Morphologists
• Surgical Pathology Foundation of diagnostic
  pathology
• Practical diagnostic considerations such as is
  it malignant is it extirpated?
• Prognostic Variables
• Tumor Grade
• Tumor Histologic Classification

The first personalized medicine Tumor typing
tumor grading proliferation, degree of
differentiation, necrosis, apoptosis
6
Anatomists Morphologists
Scientists The Molecular Pathologists
Anatomists Morphologists
Clinicians The Diagnostic Surgical Pathologists
7
Anatomists Morphologists
Molecular Pathologists Diagnostic Surgical
Pathologists
Discovery of Targets
Accurate Reproducible ID of Targets in Patient
Samples
Identification of Prognostic Predictive Markers
for Modern Medicine
8
Breast Pathology Today

• Diagnostics
• Prognostics
• Prediction of Therapeutic Response
• Tools
• Routine HE Sections Frozen Fixed
• Immunohistochemistry
• Molecular Techniques DNA RNA

Suffer from Preanalytic Analytic Sources of
Variability
9
(No Transcript)
10
Morphologic Tools for the FutureHE IHC Slide
PreScreening

• Digital imaging platforms with image analysis
  software that use slide barcodes to run block
  specific algorithms to assess for
• Presence of tumor
• Tumor size
• Distance from tumor to margins
• Presence of metastatic tumor within SLN
• Keratin stains?
• Presence of lymphovascular invasion
• IHC markers?

11
Classifier Development for Finding Tumor An
Example
12
(No Transcript)
13
Histologic Grade
14
Morphologic Tools for the FutureHE IHC Slide
PreScreening

• Digital imaging platforms with image analysis
  software that uses barcodes to run block specific
  algorithms to assess for
• Degree of tubule formation
• Mitotic counts controlling for cell density
• Nuclear anaplasia (the least reproducible in
  current Nottingham grading system)
• ? apoptotic figures, necrosis, microvessel
  density

15
Classifier Development for Quantitating Tumor
Cell Number An Example
16
Prognostic Predictive Markers ER HER2
Receptors
ER
HER2
17
Estrogen QuantitationAllred Score Continuous
Variable
Harvey JM, et al. J Clin Oncol 1999, 171474-81
18
Hormone Receptor TestingContinuous Semi-Variable
or Dichotomous?
Can this be done by computer assisted algorithms
in a more reproducible manner?
N825 cases
Collins LC, et al. Am J Clin Pathol 2005,
12316-20
19
Positive (3) Strong, membranous (chicken
wire) signal in gt30 of tumor cells.
20
Kaplan-Meier Estimates of Disease-free Survival
(Panel A) and Overall Survival (Panel B)
The addition of Herceptin in the adjuvant setting
is associated with a 16 difference in disease
free survival
Can this be done by computer assisted algorithms
in a more reproducible manner?
Romond, E. et al. N Engl J Med 20053531673-1684
21
Pathology Today

• Molecular Medicine Technologies to more
  accurately reproducibly quantitate DNA RNA
• Drive development of Prognostic Predictive
  tools
• Hormone Therapy - cheap
• Cytotoxic Chemotherapy relatively cheap
• Targeted Therapies - costly

Reduce or Obviate Preanalytic Analytic Sources
of Variability
22
Two RNA Analysis MethodsPCR DNA Arrays
Q-RT-PCR Assay (TaqMan)
DNA Arrays (Chips)

• Fixed paraffin or unfixed tissue
• 768 genes/30 mm
• Wide dynamic range (thousands fold)
• High sensitivity, specificity, reproducibility

• Unfixed or recently fixed tissue
• 1000s of genes
• Limited dynamic range (100 fold)
• Difficult to control

Cronin et al. Am J Pathol. 200416435-42. Karsten
et al. Nucleic Acids Res. 2002 30(2)E4.
23
Continuous Measurement of ER and PRN-, ER Pts
Tam vs Placebo
Overall Range of PR 1000-fold
Range of ER for ER 200-fold
Overall Range of ER 3000-fold
SD lt 0.5 Units
24
Quantitative ER Score is a strong predictor of
tamoxifen benefit

• Please note
• The magnitude of benefit of tamoxifen is already
  included in the Average Rate of Distant
  Recurrence associated with the Recurrence Score
• This study reflects only the effect of tamoxifen
  on distant recurrence rate and does not include
  other benefits such as reduction in contralateral
  breast cancer recurrence

This method does not suffer from preanalytic
analytic sources of variability
Paik S et al, 41st Annual Meeting of the American
Society of Clinical Oncology, May 13-17, 2005
Orlando, FL. Abstract 510.
25
Unmet Clinical Need for Better Markers
High risk
Biopsy or Resection
Robust markers
Low risk
26
Unmet Clinical Need for Better Markers
High risk/Large chemo benefit
Optimize chemotherapy local therapy hormonal
therapy
Biopsy or Resection
Robust markers
Optimize local therapy and hormonal therapy
Low risk/Little chemo benefit
27
21-Gene Recurrence Score (RS) Assay(Oncotype DX)
16 Cancer and 5 Reference Genes From 3 Studies
PROLIFERATION Ki-67 STK15 Survivin Cyclin B1 MYBL2
ESTROGEN ER PR Bcl2 SCUBE2
BAG1
GSTM1
INVASION Stromelysin 3 Cathepsin L2
CD68
REFERENCE Beta-actin GAPDH RPLPO GUS TFRC
HER2 GRB7 HER2
Paik et al. N Engl J Med. 20043512817-2826.
28
Standardized Quantitative RT-PCR 21 Gene Assay

• Recurrence Score in N-, ER patients

Lower RSs Lower likelihood
of recurrence Greater magnitude of tamoxifen
benefit Minimal, if any, chemotherapy benefit
Higher RSs Greater likelihood
of recurrence Lower magnitude of tamoxifen
benefit Clear chemotherapy benefit
Refs 1) Paik et al NEJM 2004, 2) Habel et al
Breast Cancer Research 2006, 3) Paik et al JCO
2006, 4) Gianni et al JCO 2005
29
Process starts with the FPET block
30
Clinical Laboratory Process

• Is there tumor present?
• Is there sufficient tumor?
• Histologic subtype

31
Plate and Sample Tracking
LIMS, Reagent, and Robotic integrated bar-coding,
tracking, and Assembly

Patient FPET Sample
RNAExtraction
Reverse Transcription cDNA Plates
ReversePrimer Pool
96-well Plate
96-well Plate
QPCRMaster Mix
96-well Plate
96-well Plate
SARPGEMTools
Material Manager
Primers
OligoPlate Assembly
Pool
96-well Plate
384-well Plate
QPCR Reaction Plate
Probe
ABI 7900
Material Manager
Tecan Robotics
Plate Layout Template and Assembly for Primers
and Samples
32
QPCR Data Acquisition
Data Import Services acquire, validate, and load
data as laboratory runs complete
ABI 7900
File Processing Service
Database Repository
Quantitation, run profile,kinetic data
Data stored locally during run
Service automatically archives and imports data
from each 7900
Data Export
File Archive
Barcode.TXT Barcode.SDS (Well.TXT)
33
Report Generation and Approval

• Result and Failure Reports
• Electronic PDF Format
• XML-Based Content Generation
• Optimized for Print, Fax, Online
• Reviewed Online by CLS
• Electronic Signature

34
Digital PathologyGHI Implementation Timeline
late/2008 Phase II Implementation
6/2008 Aperio System Purchased
7/2008 Phase I Implementation
late/2009 Phase III Implementation
35
Clinical Applications Phase I

• Has not supplanted review of HE slide
• Problem slides are uploaded prior to calling
  referring pathologist for case discussion
• Slides are scanned and uploaded for expert review
  by breast expert

36
Clinical Applications Phase II-III

• All HE slides scanned for review archival
  study reviews
• Digital image assessed in house or outside
• Slides marked for microdissection by in house or
  by submitting pathologist
• Dissection lines can be transferred from digital
  images to unstained slides for manual
  microdissection by technologists
• Algorithms for tumor identification?

Digital Pathology Cockpit
37
Atypical medullary carcinoma have highest
proliferation levels whereas tubular have the
lowest
38
RD Applications Today
39
RD Applications Phase ITumor Heterogeneity

• Tumor blocks from 8 patients

RT-PCR Analysis
1) 600 um cores taken 2) Whole sections
taken
Core 0 Core 1 Core 2 Core
3 Whole Section 1 Whole Section 2
Day 1 Later
Day Later Day
40
RD Applications Phase ITumor Heterogeneity
Within Patient Heterogeneity
Between Patient Heterogeneity
41
RD Applications Phase ITumor Heterogeneity
Within Patient Heterogeneity
Between Patient Heterogeneity
42
Reproducibility of RT-PCR Results in Cores and
Whole Sections - RS
Poorly Sampled Case
104 3 47 51 63 71
80 87
Case
43
Poor Sampling Case 104C

• Digital imaging allows for correlation of
  morphology gene expression profiles
  simultaneously
• Use of image analysis software allows for cell
  IHC quantitation and then direct correlation with
  gene expression profiles

Tumor Poor Regions
44
RD Applications Phase II-III

• Biomaterial Repository
• All slides scanned upon receipt to tissue bank
• Web browser All slides genomic profiles, DNA
  sequencing, methylation profiles, IHC data,
  morphology clinical data associated with image
• Clinical Trials
• Repository of single HE slide for multiple users
• Repository of semiquantitative IHC data
• Repository of corresponding gene expression data

45
RD Applications Phase IIIntegrated Biomaterial
Repository
46
Conclusions

• Digital Pathology likely has immediate utility in
  the surgical pathology of breast cancer
• IHC interpretation
• Future margin assessment, tumor id, image
  analysis
• Use at GHI
• Clinical use
• Discussion of problem cases
• Review of cases
• Use in retrospective studies
• RE uses
• Fulcrum of our BMR integrating all genomic,
  expression and protein data sets with digital
  images of tumors

47
(No Transcript)