Mogens Vyberg & Søren Nielsen NordiQC Institute of Pathology Aalborg University Hospital Aalborg, Denmark

10 years of NordiQC Why are 30% of

labs still getting it wrong?

May 29th 2015

• Nothing to declare

3

• International organization for quality assurance of IHC

• Founded 2003 by Nordic pathologists

• Independent, scientific, not-for-profit organisation

• Institute of Pathology, Aalborg University Hospital, DK

• General module: 3 runs/year • 15-18 different marker challenges

• Breast cancer IHC module: 2 runs/y • HER-2, ER/PR, Ki67/E-Cad …

• HER-2 ISH module: 2 runs/year • BRISH, FISH

• www.nordiqc.org

Nordic immunohistochemical Quality Control

4

WW

W.N

OR

DIQ

C.O

RG

5

6

WW

W.N

OR

DIQ

C.O

RG

Multi-tissue FFPE blocks

10% NBF 24-48 h (ASCO/CAP guidelines …)

• Normal and clinically relevant tumour tissues

• Different levels of antigen expression

• high, moderate, low, none

Test material

2 unstained slides for each marker send to the participants

1 stained slide returned for central assessment

8

The slide to be stained for Bcl-6 comprised: 1. Tonsil, 24 h.

2. Tonsil, 48 h.

3. Follicular lymphoma, grade I

4. Follicular lymphoma, grade II

5. Diffuse large B-cell lymphoma

Test material

HE

LE

NE

• Tissue selection • High • Low • None

Expressor

Nordic immunohistochemical Quality Control

0

100

200

300

400

500

600

700

800

900

2009 2010 2011 2012 2013 2014 2015

Participants

~ 90 Markers in

NordiQC runs

Tested

1-15 times

NordiQC test - haempath

1x 2x 3x 4x 5x

CD8, CD14, CD19, CD163,

IgL, MUM1 X

ALK, BSAP, Bcl2, CD45,

CD56, CD138, p53, TdT X

Bcl6, CD4, CD20, CD34,

CD68, CD79a, IgK, Ki67 X

CD3, CD10, CD23, CD30,

Cyclin D1, IgM X

CD5, CD15 X

35%

33%

21%

11%

Optimal

Good

Borderline

Poor

NordiQC assessment results 2003 – 2014

General module ~ 20,000 slides ( ~100.000 core sections)

Insufficient 32%

58% 21%

9% 12%

Optimal

Good

Borderline

Poor

NordiQC assessment results 2003 – 2014

Breast cancer module ~ 9,000 slides (~35,000 core sections)

Insufficient 21%

Publications

AIMM 2014, 22:449

AIMM 2013, 21:64

AIMM 2011, 19:437

AJCP 2005,124:782

Publications

AIMM 2015, 23:1

AIMM 2014, 22:241

Lab. A Lab. B

Serial sections stained for Estrogen receptor

ER in ductal breast carcinoma

Serial sections stained for Estrogen receptor

Lab. A Lab. B

False neg.

19

Serial sections stained for Estrogen receptor

Control: uterine cervix

Lab. A Lab. B

False neg.

20

Serial sections stained for Estrogen receptor

Control: uterine cervix

Clone 6F11 in 15/37 labs Clone SP1/EP1/1D5 in 225 labs

False pos.

21

21

NordiQC runs for HER2 IHC

CK7

Ampl. 3+ Ampl. 2+ Unampl. 2+ Unampl. 0

Optim

al

Ampl. 3+ Ampl. 1+ Unampl. 1+ Unampl. 0

Poor

22

22

NordiQC runs for HER2 IHC

CK7

Ampl. 3+ Ampl. 2+ Unampl. 2+ Unampl. 0

Optim

al

Ampl. 3+ Ampl. 2+ Unampl. 3+ Unampl. 1

Poor

23

Optimal: 36%

Good: 33%

Borderline:

Poor:

too weak / false neg.: ~ 90%

over-stained / false pos.: ~ 10% { 31% }

NordiQC general results

24

Major causes of insufficient stains in ~ 9,000 slides

Less successful antibodies/RTUs 17 %

Inappropriate antibody dilution 20 %

Inappropriate epitope retrieval 27 %

Inappropriate detection kit 19 %

Other inappropriate lab. performance 17 %

Endogenous biotin reaction (EBR)

Section drying-out after HIER

Technical platform error

. . . .

Unexplained

NordiQC general results 2003 – 2013

25

Less successful antibodies 17 %

Poor antibodies

Poor ready-to-use formats

Less robust antibodies

Platform dependent antibodies

Other error-prone antibodies

• Lot-to-lot variation

• Mouse-anti-Golgi (MAG) reaction

• Poor cocktail composition

• ….

NordiQC general results

NordiQC regrets any offence caused to laboratories and companies …

26

Poor antibodies

27

Poor antibodies (few examples)

Antigen Clone High

expressor

Low

expressor

Non

expressor

CD5 CD5/54/F6 √ FN –

CD23 MHM6 √ FN –

CD31 1A10 (√) FN –

CD31 SP38 * (√) FN –

CD138 5F7 (√) FN –

CDX2 SP54 * (√) FN FP

CDX2 CDX2-88 √ FN FP

CEA TF-3H8-1 √ √ FP

CGA DAK. A3 √ FN –

PR SP2 * √ √ FP

SYP SY38 √ FN –

28

Poor antibodies: CD5

CD5 N Sufficient* Optimal*

4C7 conc 145 74% 49%

SP19 conc 11 91% 46%

CD5/54/F6 conc 28 4% 0%

* With optimal protocol settings

29

Poor antibodies: CD5

SP19 CD5/54/F6

Tonsil

B-C

LL

TP FN

FN TP

Optimal (16%)

Poor antibodies: CD31

JC70A 1A10

Optimal (16%)

Poor antibodies: CD31

JC70A 1A10

Optimal (16%)

Poor antibodies: CD31

JC70A 1A10

Haemangiosarcoma

NordiQC performance

Not Dako Dako

Poor antibodies – MLH1

MLH1 clone ES05 MLH1 clone EPR3894

35

Poor RTU formats

Replace laboratory developed assays

Reduce workload to calibrate and validate IHC assays

Reduce need for technical competences in IHC

Optimize workflow / reduce protocol variables

Add value to inter- and intra-laboratory consistency

Requires a correctly calibrated system

Requires precise information on protocol and control

Ready-To-Use system potential:

NordiQC performance

NordiQC performance

Proportion of protocols based on concentrates vs. RTU formats in NordiQC

AMACR CD10 CK LMW CK HMW MLA

BCL2 BCL6 BSAP CD99 EMA WT1

BCL6 CD15 GLP3 MLA PAX8

NordiQC assessment

760-2505: HIER

CD45 RTU format Vendor protocol Off-label protocol

Epitope retrieval None HIER CC1, 64 min.

Primary Ab inc. 16 min. 16 min.

Detection system 2-step multimer 3-step multimer

Control Tonsil Tonsil + liver

Reaction pattern Lymphocytes Lymphocytes + macroph.

Assessment score Borderline – False neg. Optimal

Tonsil + liver

Plasmacytoma

CD5 Run 24 N Sufficient* Optimal*

SP19 conc 11 91% 46%

SP19 RTU Dako 3 100% 100%

SP19 RTU VMS 14 79% 14%

Poor RTU formats: CD5

* With optimal protocol settings

FN

CD5 Run 34 N Sufficient* Optimal*

SP19 RTU VMS 33 97% 97%

Poor RTU formats: CGA

LK2H10 REF

pAb RTU Company 1

mAb LK2H10 RTU Company 3

mAb LK2H10 RTU Company 2

Medullary carcinoma

Poor RTU formats: CGA

Small cell carcinoma

LK2H10 REF

pAb RTU Company 1

mAb LK2H10 RTU Company 3

mAb LK2H10 RTU Company 2

42

Platform dependant antibodies

43

Platform dependent antibodies Antigen Clone XT / Ultra

automated

Bond-max automated

Autostainer semiautomated

CD4 1F6 FN Weak √

SP35 √ √ √

CD56 123C3 FN Weak √

MRQ-42 √ ? √

CD79a JCB117 Weak √ √

SP18 √ √ √

BSAP/Pax5 24 FN Weak √

SP34 √ √ √

BCL6 PG-B6p FN Weak √

GI191E/A8 √ √ √

SYP 27G12 Weak √ √

MRQ-40 √ √ √

Hodgkin lymphoma NS

clone SP34 RTU VMS/CM

clone 24 RTU VMS/CM

Platform dependent antibodies: PAX5

x200 x200

45

Inappropriate antibody dilution

Inappropriate antibody dilution – Ig light chains

~1:300 ~1:3.000 ~1:30.000

IgK: Dako pAb A0191

Inappropriate antibody dilution – Ig light chains

239 IgK tests, 12 Abs:

12% optimal

Dako pAb A0191:

17% optimal

+TRS/Ci 3.000-16.000:

29 % optimal

All other Abs:

0% optimal

48

Inappropriate visualization system

NordiQC run 41/42 2014 - MMR

MMR MLH1 mAb clone ES05, 1:20 Leica

UltraView + Amplification OptiView + Amplification (Tyr.)

1’ generation 3-step multimer, VMS 2’ generation 3-step multimer, VMS

NordiQC run 41 2014 – PMS2 131 labs

NO

mu

tation

Optimal: 47% Insufficient: 15%

NordiQC run 41 2014 – PMS2 131 labs

Mu

tation

Optimal: 47% Insufficient: 15%

NordiQC run 41 2014 – PMS2 131 labs

Mu

tation

Optimal: 47% Insufficient: 15%

•Too dilute Ab •Insufficient HIER •Insensitive viz system

53

Inappropriate epitope retrieval

&

Misleading data sheets

54

Inappropriate retrieval (31%)

AE1/AE3 + HIER AE1/AE3 + proteolysis

Liv

er

RCC

FN TP

FN TP

55

Misleading datasheets

Giving false negative results when only LMW-CKs are present

Misleading datasheets – improved information

57

IHC - NordiQC 2014

AE1/AE3 : Optimal results only obtained by HIER in NordiQC runs

Dako: RTU – HIER Conc: Proteolysis or HIER Leica: RTU – Proteolysis Conc: HIER Thermo: Conc: HIER Quanto – Proteolysis UltraVision ………… VMS: RTU - Proteolysis Misleading data sheets + Wrong control material used

By 17th October 2014

Improved datasheets

NordiQC run 41 2014 – ECAD 271 labs

False positive: EP700Y

Fra: Galloway, Mary [mailto:Mary.Galloway@fda.hhs.gov] Sendt: 13. november 2014 01:14 Til: Søren Nielsen / Region Nordjylland

Emne: RE: Changes Made to Package Inserts

Sören,

Thanks for identifying and alerting us to the issues with … anti-Pan Keratin.

The package inserts are now changed (see links below).

I hope we can continue to learn of any future staining problems you may

uncover.

Much appreciated!

Mary

RCC

FN TP

60

Misleading datasheets Antigen Clone Company Datasheet Result

CGA Lk2H10 VMS No retrieval FN

CK8 5D3 Leica RTU: HIER

Conc: proteolysis

Confound

FN

CK19 RCK108 BioGenex Proteolysis FN

CK19 B170 Leica Proteolysis FN

CKPan AE1/AE3 VMS/Dako Proteolysis FN

CD34 QBEnd 10 Leica RTU: HIER

Conc: proteolysis

Confound

FN

CD34 QBEnd 10 VMS Changed from no

retrieval to HIER

FN /

OK

CD68 KP1 Thermo Proteolysis FN

DES DE-R-11 Cell Marque Proteolysis FN

PLAP PL8-F6 BioGenex No retrieval FN

VIM 3B4 VMS Proteolysis FN

WT1 6F-H2 Dako RTU: HIER

Conc: proteolysis

Confound

FN

Misleading datasheets

Giving false negative results in low expressing cells and tumours

62

Tailored NordiQC recommendations

63

Tailored recommendations

Replace less successful antibodies (conc./RTU)

Calibrate the antibody concentration

Use HIER (instead of proteolysis or no retrieval)

Increase HIER time / temperature / buffer pH

For 95% of epitopes pH 8-9 is preferable to pH 6

Use a non-biotin based viz. system

Use FDA approved kits instead of home-brews

. . . . .

Improve the internal QC: Identify the right controls –

Select well defined normal low expressor cells/tissues

64

419 advices for 11 markers

No. Improved %

Positive 268 195 73

Negative 151 21 14

Results of NordiQC recommendations

NordiQC EQA: Estrogen Receptor 2003-11

0

10

20

30

40

50

60

70

80

90

100

8 10 13 B1 B3 B5 B7 B8 B10 B11 B13 B15 B17

PASS RATE (%)

45%

87%

NordiQC EQA: Estrogen Receptor 2003-11

0

10

20

30

40

50

60

70

80

90

100

8 10 13 B1 B3 B5 B7 B8 B10 B11 B13 B15 B17

PASS RATE (%)

70

281 122

141

197

Participants

45%

87%

HER-2 staining results in 17 runs

Roche – NordiQC joint venture

”Normalized” to the American breast cancer population

*

* The large majority due to 1+ reactions in amplified 2+ tumours

Roche – NordiQC joint venture

”Normalized” to the American breast cancer population

Roche – NordiQC joint venture

For each 1$ saved by the pathology lab by usage of cheaper reagents, the healthcare system is ultimately burdened with ~ 6$

Immunohistochemical expression of HER2 in breast cancer:

Socioeconomic impact of inaccurate tests

Vyberg M, Nielsen S, Røge R,

Sheppard B, Ranger-Moore J, Walk E, Gartemann J, Rohr UP, Teichgräber V

NordiQC, Aalborg, DK, Ventana Medical Systems Inc, Tucson, AZ ,

F. Hoffmann-La Roche Ltd, Basel, Switzerland

Submitted for publication

Almost 1/3 of all IHC stains produced by NordiQC

participants are still insufficient !

New labs

New antibodies, techniques, platforms

Increasing demands

How many IHC stains produced by labs not

participating in an EQA scheme are insufficient ?

How many scientific publications are based on

insufficient IHC stains ?

What are the consequences for the patients ?

Perspective

72

External Quality Assurance (EQA)

Provides objective evidence of lab performance

Identifies methodological errors

Provides directions for improvements & controls

The results of the NordiQC work indicate that

Improvement of IHC is strongly needed

EQA schemes, industry and KOL must align - describing the requirements for optimal IHC performance.

Conclusion

73

74

Thank you for your attention!