Quantitative Assessment of Tissue- based IHC Biomarkers Next Generation Pharmaceutical Summit David Young 7 Apr 09.

Quantitative Assessment of Tissue-based IHC Biomarkers

Next Generation Pharmaceutical Summit

David Young 7 Apr 09

2Digital Pathology

Digital Pathology – Research and Clinical Possibilities

Quantitative Digital pathology

IHC – Traditional evaluation vs Image analysis

Tools not limited to pathologists

Digital Pathology – Where Are We Headed?

4Digital Pathology

Digital Pathology – Research and Clinical Possibilities

–Archival of pathology specimens

–Diagnosis

–Digital slide conferencing

–Consultation

Help from Development Teams

– putting the power in the hands of the people who know it best

Quantitative Digital Pathology - The Next Step

6Quantitative Digital Pathology

Pathologist opinions

–Good enough for government work, or

–Close, but no cigar

X number of pathologists = Y number of results

–Diagnoses

–IHC analysissubjective; based on familiarity of tissue

and experience

IHC Assessment of Tissue-based Biomarkers

8Immunohistochemistry Analyses and Quantitative Digital Pathology

Not an exact science

Basis of many aspects of drug development and drug selection

9

Biomarker Scoring Consensus

Clark (2006) – ‘there is no consensus in the literature about how to summarize these scoring assessments into a single determination of EGFR protein expression status as EGFR positive or EGFR negative.’

‘Evaluation of the clinical significance of EGFR expression by IHC has been complicated by the use of different antibodies, different scoring systems, and different clinical endpoints.’

Clark, et al: J Thorac Oncol 2006

10

Prevalence and tumor surveillance

Prognostic factors

Predictive factors

Comparing study results from a recognized baseline of analysis

Importance of Standardized Scoring

11IHC Scoring Concordance – Pathologists Variability

Colon

0

50

100

150

200

250

300

0 100 200 300

Lung

0

50

100

150

200

250

300

0 100 200 300

ConcordanceTotal scoring = 78%Cut point <100 = 92%

ConcordanceTotal scoring = 75%Cut point <100 = 100%

12Pathologist Variation

Legend:

Red – Pathologist 1

Blue – Pathologist 2

0

50

100

150

200

250

300

0 50 100 150 200 250 300

Assay B Score

Ass

ay A

Sco

re

Pathologist 1 Scores:

Y = 0.96X + 3.21

R = 0.987

Pathologist 2 Scores:

Y = 0.97X -2.72

R = 0.974

13Image Analysis – Lessens Subjectivity of Scoring

Quantify:

Size (area)

Positive cells

Negative cells

Intensity levels

14

E-Cadherin

–Marker of epithelial phenotype

–Associated with cell-to-cell adhesion

–Membrane protein

Vimentin

–Marker of mesenchymal phenotype

–Associated with cellular skeleton

–Cytoplasmic protein

Tissue-based Biomarkers – Case Study

15Experimental Xenograft model

H&E E-cad Vim

16Heterogeneity in Tumor Tissue – E-cad

17Heterogeneity in Tumor Tissue – Vim

18

Traditional IHC Score (H-Score)

IntensityScore (IS) 1 = weak0 = negative 2 = intermed 3 = strong

0 – 100%ProportionScore (PS)

100%75%30%10%1%0

Score range: 0-300

19Factors Affecting IHC Analysis – Not Just the Pathologist

Tumor acquisition (pre-analytical factors)

Tumor size

Tumor type (Tumor tissue and host response)

Antibodies

Processing factors

Individual variation in evaluation

20Cell Culture - E-cadherin

Algorithm - Membrane v9 Default  

Min Nuclear Size (um^2) 10 85

Background Intensity Threshold 240  

Weak (1+) Intensity Threshold 200  

Moderate (2+) Intensity Threshold 170  

Strong (3+) Intensity Threshold 105  

21NSCLC Criteria setup

22Cell Culture - Vimentin

Algorithm - Color Deconvolution v9 Default

Weak Postive Threshold 220 180

Medium Postive Threshold 175 120

Strong Positive Threshold 100 60

23Xenograft model - E-cadherin

Entire Specimen IHC Test box

(3+) Percent Cells 71.83 50 65.67

(2+) Percent Cells 9.61 40 8.17

(1+) Percent Cells 18.53 10 26.16

(0+) Percent Cells 0.03 0 0.00

SCORE 253.24 240 239.51

24Xenograft model - Vimentin

Entire Specimen IHC Test box

(3+) Percent 41.28 25 24.70

(2+) Percent 46.02 50 56.80

(1+) Percent 20 20 17.51

(0+) Percent 0.26 5 1.00

SCORE 228.32 195 205.21

25NSCLC – example 1

26NSCLC – example 1 (higher mag)

E-Cad Vim

Aperio IHC Aperio IHC

(3+) Percent Cells 41.28 25 (3+) Percent 1.10 1

(2+) Percent Cells 46.02 50 (2+) Percent 0.96 1

(1+) Percent Cells 20 20 (1+) Percent 3.03 3

(0+) Percent Cells 0.26 5 (0+) Percent 94.90 95

SCORE 228.32 195 SCORE 8.25 8

27NSCLC – example 2

E-Cad Vim

Aperio IHC Aperio IHC

(3+) Percent cells 61.77 35 (3+) Percent 6.93 0

(2+) Percent cells 17.73 60 (2+) Percent 30.51 10

(1+) Percent cells 20.00 5 (1+) Percent 43.15 30

(0+) Percent cells 0.50 0 (0+) Percent 19.41 60

SCORE 240.77 230 SCORE 124.96 50

28NSCLC – example 3

E-Cad Vim

Aperio IHC Aperio IHC

(3+) Percent Cells 65.28 15 (3+) Percent 3.80 0

(2+) Percent Cells 9.88 15 (2+) Percent 2.21 10

(1+) Percent Cells 24.80 50 (1+) Percent 6.05 15

(0+) Percent Cells 0.04 20 (0+) Percent 87.94 75

SCORE 240.40 125 SCORE 21.87 35

29NSCLC – example 4 (Whole tumor; E-Cadherin)

E-Cad

Aperio IHC

(3+) Percent Cells 68.60 50

(2+) Percent Cells 6.25 25

(1+) Percent Cells 24.54 20

(0+) Percent Cells 0.60 5

SCORE 242.84 220

30NSCLC – example 4 (Vimentin)

Vim

Aperio IHC

(3+) Percent 0.44 0

(2+) Percent 0.96 0

(1+) Percent 2.50 0

(0+) Percent 96.11 100

SCORE 5.74 0

31Pancreas – Xenograft 1

H&E E-cad Vim

32Pancreas – Xenograft 1

E-Cad Vim

Aperio IHC Aperio IHC

(3+) Percent Cells 71.59 50 (3+) Percent 0.40 0

(2+) Percent Cells 7.48 40 (2+) Percent 1.26 0

(1+) Percent Cells 20.93 10 (1+) Percent 24.36 0

(0+) Percent Cells 0 0 (0+) Percent 73.98 100

SCORE 250.66 240 SCORE 28.08 0

33Pancreas – Xenograft 2

E-Cad Vim

Aperio box

Aperio whole IHC

Aperio box

Aperio whole IHC

(3+) Percent Cells 0 0.76 0 (3+) Percent 50.82 51.80 70

(2+) Percent Cells 0 10.65 0 (2+) Percent 34.35 34.72 30

(1+) Percent Cells 22.22 49.27 0 (1+) Percent 13.26 12.42 0

(0+) Percent Cells 77.78 39.32 100 (0+) Percent 1.58 1.06 0

SCORE 22.22 72.85 0 SCORE 234.42 237.26 270

34Summary – What have we learned so far?

Selection of site for IHC evaluation is important; may or may not be reflective of whole tumor

Tumor heterogeneity affects tissue-based biomarker assessment and analysis

IA correlates well with traditional IHC scoring methods.

Validation removes pathologists scoring variability

‘Tweaking’ of algorithms required prior to universal deployment

Putting the Power in the Hands of the People

36Investigator Asks the Questions

Thank you!