Intrinsic and Extrinsic Controls for Formalin Fixed, Paraffin ......TN98 TN61 TN41 TN81 TN93 TN80 TN73 TN37 TN106 TN52 TN101 TN71 TN88 TN30 TN16 TN43 TN86 TN100 TN50 TN48 TN29 TN25

Intrinsic and Extrinsic Controls for Formalin Fixed, Paraffin Embedded Tissue

David L. Rimm M.D., Ph.DDirector, Yale Pathology Tissue Services

Professor, Dept. of PathologyYale University School of Medicine

Disclosure/Disclaimer• I am a consultant to, stockholder in, and scientific co-founder of

HistoRx Inc. the exclusive licensee of the AQUA® technology

• I am an author on the Yale held patent on the AQUA technology and receive royalties.

• This project has been funded in whole or in part with the federal funds from the National Cancer Institute, National Institutes of Health, under Contract No. HHSN261200800001E. The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or organizations imply endorsement by the U.S. Government.

The central problem: Standardization of Protein

Assessment of Formalin-Fixed Paraffin-Embedded tissue

• Definition of Extrinsic vs Intrinsic Control• Solution of the Extrinsic control problem• Progress toward an Intrinsic control or

Tissue Quality Index (previously designated TIC for tissue immuno-competence index)

Extrinsic vs Intrinsic Controls

• Extrinsic controls; control for and standardize all the processes from the stainer through the analysis

• Intrinsic controls; control for and standardize all the processes from the patient to the stainer (pre-analytic varibles)

Our solution to the Extrinsic Control Standardization Problem: The AQUA method of

Quantitative ImmunofluorescenceAQUA: Think like a moleculeSelection of regions only as a function of colocalization of molecular interactions

Other Software: Think like a human Assign significance to morphologically defined entities and use feature extraction to emulate human assignments

Example: a nuclear protein is measured by colocalization with DAPI in a cytokeratin positive region

Example: a nuclear protein emulates the human definition of nucleus and finds round or roundish entities, then counts signal within the roundish entities

http://www.tissuestudio.com/

http://www.tissuestudio.com/�

Understanding the Difference between AQUA and other tissue analysis software

AQUA: Think like a moleculeSelection of regions only as a function of colocalization of molecular interactions

Other Software: Think like a human Assign significance to morphologically defined entities and use feature extraction to emulate human assignments

Solution: No pathologist to “agree” with since result is strictly derived from co-localization

Problem: Feature extraction software does not “agree” with the pathologist since tumors (and pathologists) are very different

http://www.tissuestudio.com/

0

2

4

6

8

10

12

CookieM Animal Beaker CookieM CookieM CookieM CookieM

CW JM JM JM MC MC MC

Day 1 Day 1 Day 1 Day 1 Day 1 Day 2 Day 3

AQ

UA

(R)

Sco

re (

log2

)

PM2PM1 PM1PM3 PM1

PM1 PM1

CV=4.3%

http://www.tissuestudio.com/�

TMA

WTS

TMA-Tissue MicroarrayWTS-Whole Tissue Section

Cytokeratin Tumor Mask

Combine DAPI image and cytokeratin image then cluster to assign each pixel to a subcellular compartment

Estrogen Receptor

Σ compartmentpixel area

Σ target intensityin compartment pixels

= AQUAscore

Generating the AQUA®

score

40 patient controls(range of ER)

formalin-fixing &paraffin-embedding

lysates

Western Blot alongside

recombinant ER

calculate ER as AQUA scores

Convert AQUA scores to pg/ug or other units

Control TMA

cell line panel(known range of ER)

calculate ER in pg/µg

Alley Welsh

Standardized Index Array ER antibody used is 1D5

Lowest positive vs. highest negative

ER

ER Alley Welsh

Expanded “levels” to visualize thresholdcontracted dynamic range of grayscale (max RGB input level 25516)

ER ER

ER ER

HC

C22

79H

2882

H19

3H

441

H12

99H

1666

H13

55S

KB

R3

MC

F7 10 n

g25

ng

50 n

g12

5 ng

EGFR

Rec EGFR

Beta tubulin MCF7

H1299 H1355

H193

H2279

H1666H2282H441

EG

FR le

vels

(ng/µg

)

Aqua score

SKBR3

0.85 ng/µg EG

FR le

vels

(ng/µg

)

EGFR negative EGFR positive

0.85 ng/µg

A.B.

C. D.

50 KD

91 KD

170 KD

EG

FR le

vels

(ng/µg

)

R=0.782

25

20

15

10

5

0

HC

C2279

H193

H1666

H2882

H441

H1299

H1355

SK

BR

3

MC

F7

12

10

8

6

4

2

0

109876543210

20 40 60 80 100 120 140 160

Tassos Dimou

Development and Commercialization Of a Quantitative Protein Measurement Technology

(AQUA) from the lab to the patient

Extrinsic vs Intrinsic Controls

• Extrinsic controls control for and standardize all the processes from the stainer through the analysis

• Intrinsic controls control for and standardize all the processes from the patient to the stainer (Pre-analytic variables)

Goals of our Project• Development of a Tissue Quality Index

(TQI): • obtained by developing a quantitative

intrinsic control that can measure the degree of degradation of any FFPE sample.

Pre-analytical Variables (incomplete list)

• Variable warm ischemic time• Variable cold ischemic time• Variable manipulation during gross cutting and prepping• Variable temperature during fixation• Variable total fixation time• Variable thickness of tissue blocks• Variable half life of fixative• Variable types/brands/ components of fixatives• Variable types of tissue processors• Variable solutions in the processor• Variable temperatures of different processor components• Variable types of embedding paraffin• Variable slide drying times• Variable slide oven temperatures

Pre-Analytic Variables; Can we treat them as a black box?

If we cannot control pre-analytical variables can we

quantify the damage or tissue degradation caused by them?

Can we disqualify specimens for companion dx testing?

Goal 1: To generate two “discovery” tissue sets to assess “pre-analytical” variability.Goal 2: Assessment of markers of cold ischemia (“housekeeping markers”) on discovery cohortsGoal 3: Assessment of markers of hypoxia on discovery cohortsGoal 4: Generation of a Multiplexed “Tissue Immunologic Competence” (TIC) Model (now Tissue Quality Index (TQI)) for normalization of tissue handling that measures tissue integrity for immunological assessmentGoal 5: Validation testing of the TIC Model in two core vs. resection specimen studies

Intrinsic Controls for FFPE tissue

ApproachGenerate

Intrinsic Control cohorts

Select and validate potential

antibodies/reagents

Test each reagent individually on

Intrinsic Control Cohorts

Generate simplest Multi-variable model

that can assess tissue quality (TQI)

Validate TQI

Perform Western Blotting on cell line positive and negative controls

Is there a band at the expected molecular weight?

Titer antibody on TMA containing control tissues and cell lines

Is staining localized and specific consistent with protein’s biological description?

Does expression level by IHC correlate with

WB?

Is the antibody reproducible between runs?

No Aband

on Antibo

dyYes

Yes

Yes

Yes

No

No

Yes

No

Antibody Validated

Histone 4 - Nuclear R =-0.398p = <0.0001

N = 77

02000400060008000

100001200014000160001800020000

0 100 200 300 400 500Time to Fixation (Mins)

His

tone

4 A

vera

ge N

ucle

ar A

qua

Scor

es

0

60

120

180

240

300

360

420

480

TN85

TN74

TN28

TN35

TN65

TN66

TN84

TN32

TN11

TN55

TN15

TN95

TN98

TN61

TN41

TN81

TN93

TN80

TN73

TN37

TN106

TN52

TN101

TN71

TN88

TN30

TN16

TN43

TN86

TN100

TN50

TN48

TN29

TN25

TN102

TN36

TN103

TN77

TN17

TN1

TN9

TN5

TN21

TN13

TN10

TN23

TN2

Tim

e to

Fix

atio

n (m

inut

es)

Construction of the Rochester Tissue Microarray (2x redundancy)

Two fold redundancyN=125 , tumor=93, normal=2, cell lines=10 control breast tumor=10 ,control lung tumor = 10Collected by Dr. David Hicks and colleague, University of Rochester Medical Center

Bx-

TRPa

ir 1

Bx

Bx-

TRPa

ir 1

TR

Tissue sample

Area of concern

Field of View (FOV)

Tumor

Core Bx – Resection Pair Cohorts

Example of CNB – Resection Cohort (from other studies)

pTyr1248HER2

1-N 2-N 3-N 4-N14-N

5-2+10

-2+13

-2+15

-2+16

-2+17

-2+18

-2+19

-2+20

-2+21

-2+22

-2+23

-2+24

-2+25

-2+9-3+11

-3+27

-3+28

-3+29

-3+30

-3+31

-3+32

-3+33

-3+34

-3+35

-3+36

-3+37

-3+38

-3+39

-3+40

-3+41

-3+0

1000

2000

3000

4000

5000

6000

CNBTRP<0.0001

Pairs

AQ

UA

Antibody Validation (Overview)




Is staining localized and specific consistent with protein’s biological

description?

Does expression level by IHC correlate with WB?


No Abandon Antibody

Yes

Yes

Yes

Yes

No

No

Yes

No

Antibody Validated

These markers show either an increase or a decrease of expression with time to fixation on the time to fixation TMA

GAPDH - Tumor Mask

R=0.076p = 0.48N = 87

0

5000

10000

15000

20000

25000

0 100 200 300 400 500

Time to Fixation (Mins)

GA

PDH

ave

rage

Tum

or M

ask

Aqu

a sc

ores

Histone 3 - Nuclear

R = 0.137p = 0.22N = 80

0

5000

10000

15000

20000

0 100 200 300 400 500


Hist

one

3 av

erag

e Nu

clea

r

Aqua

Sco

res

Beta Tubulin - Cytoplasm

R = -0.029p = 0.79 N = 85

0

5000

10000

15000

20000

0 100 200 300 400 500


Beta

Tub

ulin

Ave

rage

Cy

topl

asm

Aqu

a sc

ores

Cyclin D1-Nuclear

0

4000

8000

12000

16000

0 100 200 300 400 500Time to Fixation

Cyc

lin D

1 av

erag

e nu

clea

r A

QU

A s

core

R=-0.05p=0.66N=78

Correlation between markers of cold ischemia and hypoxia with time to fixation


N = 77

02000400060008000

100001200014000160001800020000


His

tone

4 A

vera

ge N

ucle

ar A

qua

Scor

es

YTMA 173-1-22 phospho-p44/42 MAPK Time to Fixation (Mins) one fold

R2 = 0.0233

0

2000

4000

6000

8000

10000

12000

14000


Aqu

a sc

ore

of M

APK

und

er T

M

(one

fold

)

While some show a downward or upward trend, heterogeneity is a concern

AKAP and Time to fixation (Average TM scores)

R2 = 0.0572

0

2000

4000

6000

8000

10000

12000

14000

0 50 100 150 200 250 300 350 400 450Time to Fixation (Mins)

AK

AP

av

era

ge

aq

ua

sc

ore

s

(TM

)

0

2000

4000

6000

8000

10000

12000

14000

16000

18000

20000

0 50 100 150 200 250 300 350 400 450


His

tone

4 A

vera

ge N

ucle

ar A

qua

Scor

es

Time to Fixation<1hr Aqua - 8155

Time to Fixation 1-2hrAqua - 6671

Time to Fixation> 2hr Aqua – 5102.62

Spearman Rank Correlation:

P=<0.0001

R=-0.398

N=77

Histone 4 on Time to Fixation Array

Biopsy Histone 4 nuclear

20/25

10/18

21/20

12/9

11/9

5/21

5/19

18/18

7/38

14/29

10/24 13/24

9/10

4/27

10/7

13/28

5/24

17/23

21/21

16/14

0

2000

4000

6000

8000

10000

12000

14000

sample1

sample2

sample3

sample4

sample5

sample6

sample7

sample8

sample9

sample1

0

sample1

1

sample1

2

sample1

3

sample1

4

sample1

5

sample1

6

sample1

7

sample1

8

sample1

9

sample2

0

sample2

1

sample2

2

sample2

3

sample2

4

sample2

5

BX Average H4 NucRX Average H4 Nuc

Paired t-test: p=0.88

Resection Histone 4 nuclear

Histone 4 on 25 matched pairs of biopsies and resections

Heterogeneity washes out the effect of time to fixation

phospho-MAPK test to test reproducibility

R2 = 0.9645

0

5000

10000

15000

20000

0 2000 4000 6000 8000 10000 12000 14000YTMA 179-6-116 Aqua score of MAPK under TM

YT

MA

17

9-6

-6 A

qu

a

sc

ore

of

MA

PK

un

de

r T

M

YTMA 165- Tic Reproducibility

R2 = 0.9063

02000400060008000

10000120001400016000

0 5000 10000 15000 20000 25000YTMA 165-1-25 Aqua score of MAPK under TM

YT

MA

165

-1-3

7 A

qu

a sc

ore

of

MA

PK

un

der

TM

AntibodyValidated

Assay Reproducibility

YTMA 173-1-22 phospho-p44/42 MAPK Time to Fixation (Mins) one fold

R2 = 0.0233

0

2000

4000

6000

8000

10000

12000

14000


Aqu

a sc

ore

of M

APK

und

er T

M

(one

fold

) P = .3983

The Spearman Rank Correlation shows a trend (p-value not significant) towards negative correlation.

Biopsy patient 5 pMAPK Immunohistochemistry

Resection patient 5 pMAPK Immunohistochemistry

Average AQUQ Score 5500

Average AQUQ Score 407

0

2000

4000

6000

8000

10000

12000

14000

16000

18000

20000

sample2

sample3

sample4

sample5

sample6

sample7

sample9

sample1

0

sample1

1

sample1

2

sample1

6

sample1

7

sample1

9

sample2

0

sample2

1

sample2

2

sample2

3

sample2

4

biopsy pMAPKaverage resection pMAPKaverage

Assessment of pMAPK expression on 25 matched pairs

of biopsies and resections

14/1339/11

6/14

13/31

8/15

10/28

18/12

6/79/42

18/35

14/36

12/13

23/27

21/52

6/21

44/45

30/4331/30

Samples missing: 1,8,13,14,15,18,25 Results for 18 pairs


Mixed Effects Model for Estimating Number of Fields Required for Immunostaining

Juliana Tolles, Yalai Bai and Annette Molinaro

Estrogen Receptor: Estimated Prediction Error Criterion

Model = linear regressionInput=average of sampled fields (simulated from a normal distribution for each subject)“Truth”=average over all fields


The number of FOVs required are a function of the protein examined

Number of 20X fields of view (FOVs) to find stable minimum in mixed effects modeling


Measurement of Variability and Heterogeneity of Estrogen Receptor

Between PatientsMean difference: 149 pts (CI: 92-241)

Between Biopsy and ResectionMean decrease: 134 pts (CI: 62-204)

Between Regions within a SampleMean difference: 39 pts (CI: 29-53)

Juliana Tolles and Annette Molinaro

Problem!! Need to find markers that are both highly homogeneous and highly

sensitive to pre-analytic variables

• Go beyond antibodies? Eosin• Phospho-modification? pTyr antibodies• Your suggestions here_______________

Intra Array Reproducibility of Eosin on YTMA173-2-8, march2011

R2 = 0.3405

0

500

1000

1500

2000

2500

3000

3500

0 500 1000 1500 2000 2500 3000Eosin in TM, norm, first fold

Eosi

n in

TM

, nor

m, s

ec fo

ld

Average Eosin Expression in TM on YTMA173-2-8, March2011

R2 = 0.0035

0

500

1000

1500

2000

2500

3000

3500

4000

4500

5000

0 50 100 150 200 250 300 350 400 450time to fixation in min

norm

AQ

UA

Sco

re o

f Eos

in in

TM

Distribution of norm AQUA Scores in TM for Eosin at 1 to 50, 5 min at RT

0

500

1000

1500

2000

2500

3000

3500

4000

4500

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28

Staining with Eosin at 1 to 50

-to measure the amount of free amines and protein degradation

-Predict increase free amine with increased degradation

Reproducibility for pTyrosine on breast tests

R2 = 0.6294

0

2000

4000

6000

8000

10000

12000

0 2000 4000 6000 8000 10000 12000 14000 16000 18000

first fold, pTyrosine in TM, norm AQUA Scores

sec

fold

, pTy

rosi

ne in

TM

, nor

m A

QU

A

Scor

es

AntibodyValidated

Treatment of breast test with Lambda Phosphatase for 2 hours at 37degree celsius

0

2000

4000

6000

8000

10000

12000

1 2 3 4 5 6 7 8 9 10

norm

AQ

UA

Sco

re in

TM

of p

Tyro

sine

pTyrsone1:250 phosphatase treated slide, incubated with pTyrosine 1 to 250

Sample #

Average pTyrosine Exression on YTMA173-2 and Time to Fixation

R2 = 0.0762

0

2000

4000

6000

8000

10000

12000

14000

16000

0 50 100 150 200 250 300 350 400 450time to fixation

Ave

rage

pTy

rosi

ne in

TM

, nor

m A

QU

A S

core

P=0.0079

Intra Array Reproducibility for pTyrosine 4G10 on YTMA 173-2 at 1 to 250, March 2011

y = 0.5745x + 1072.6R2 = 0.5407

0

2000

4000

6000

8000

10000

12000

0 2000 4000 6000 8000 10000 12000 14000 16000first fold pTyrosine, norm TM AQUA Scores

sec

fold

pTy

rosi

ne, n

orm

TM

AQ

UA

Sco

res

SummaryGenerate

Intrinsic Control cohorts

Select and validate potential

antibodies/reagents

Test each reagent individually on

Intrinsic Control Cohorts

Generate simplest Multi-variable model

that can assess tissue quality (TQI)

Validate TQI




Is staining localized and specific consistent with protein’s biological description?

Does expression level by IHC correlate with

WB?


No Aband

on Antibo

dyYes

Yes

Yes

Yes

No

No

Yes

No

Antibody Validated


N = 77

02000400060008000

100001200014000160001800020000


His

tone

4 A

vera

ge N

ucle

ar A

qua

Scor

es

Thanks to:Rimm Group:Valsamo (Elsa) AnagnostouBonnie Gould RothbergVeronique NeumeisterSeema AgarwalAnastasios DimouHuan ChengMaria BaqueroAlley WelshJason HannaJennifer BordeauxHalley WimberlySummar SiddiquiElisabeth RichardsonHollis VirayYalai BaiRobert Camp

Yale Pathology Tissue ServicesLori CharetteJoe Salame Aruna MadanSudha Kumar Peter Gershkovich

Outside Yale CollaboratorsKonstantinos Syrigos (Athens)Gerold Bepler (Moffitt-KCI)Daniel Hayes and SWOGElaine Alarid (UW)Bruce Haffty (CINJ)

Yale CollaboratorsAnnette Molinaro

Karen LostritoJuliana Tolles

Harriet KlugerRuth HalabanSteve AriyanDaniel Boffa

Catherine SullivanFrank DetterbeckLynn TanoueLyndsay Harris

Work supported by grants from the NCI, DOD, the Susan G Komen Foundation for the Cure and the NCI Office of Biospecimen and Biorepository Research (OBBR)

www.tissuearray.orgRimm Lab 2010

Same Genome – Different Proteome

Why we measure protein

Why we measure protein in situ

Auguste Renoir : The Luncheon of the Boating Party C.1881

Claude Monet: The Stroll, Camille Monet and Her Son Jean (Woman with a Parasol)C. 1875

Why we measure protein with a machine

47

Marker of cold ischemia: Histone H4

Western Blotting Expression Range Graph

Quantitative Immunofluorescence

0

2000

4000

6000

8000

10000

12000

14000

16000

1/50 1/100 1/500 1/1000 1/5000

bottom three Aqua scorestop three Aqua scoressubtraction

Dilution 1:50 AQUA=14307





Tubulin

H4

H441

H135

5

HCC2

279

A431

SKBR

3BT

474

MCF

-7

H166

6H1

299

MB4

68BT

20

AntibodyValidated

Histone 4 Reproducibility (Tic Array)All spots

R = 0.763

0

500

1000

1500

2000

2500

0 2000 4000 6000 8000 10000 12000

H4 Nuclear Scores YTMA 165_1_ March

H4

Nuc

lear

Aqu

a sc

ores

YTM

A 1

65_1

_ Fe

b

Assay Reproducibility

0

2000

4000

6000

8000

10000

12000

14000

16000

TN85 TN47 TN67 TN89 TN63 TN46 TN17 TN8

25min 41min 45min 49min 68min 75min 95min 135min

average pMAPK on biopsies average pMAPK on TMAspots


Assessment of pMAPK Expression on Biopsies and

matched TMA Spots on Time to Fixation Array

13

2927

15

34

30

34

20

pTyrosine Expression Range Graph

0

2000

4000

6000

8000

10000

12000

14000

16000

18000

1:25 1:100 1:250 1:500 1:1000 1:5000

Aqu

a S

core

of p

Tyro

sine

und

er T

M topbottomsubtraction

4G10 Platinum, Anti-Phosphotyrosine

Mouse monoclonal Antibody cocktail IgG2b

Millipore, Cat. # 05-1050

0

2000

4000

6000

8000

10000

12000

14000

16000

18000

1:25 1:100 1:250 1:500 1:1000 1:50000

1

2

3

4

5

6

7topbottomratio

pTyrosine Expression Range Graph

Documents

Intrinsic and Extrinsic Controls for Formalin Fixed, Paraffin ......TN98 TN61 TN41 TN81 TN93 TN80 TN73 TN37 TN106 TN52 TN101 TN71 TN88 TN30 TN16 TN43 TN86 TN100 TN50 TN48 TN29 TN25