PD‐L1 immunohistochemistryas a predictivebiomarker: · PDF filePD‐L1...

26
PD‐L1 immunohistochemistry as a predictive biomarker: pros and cons Edurne Arriola Head of Lung Cancer Medical Oncology Hospital del Mar 20/10/17

Transcript of PD‐L1 immunohistochemistryas a predictivebiomarker: · PDF filePD‐L1...

Page 1: PD‐L1 immunohistochemistryas a predictivebiomarker: · PDF filePD‐L1 immunohistochemistryas a predictivebiomarker: pros and cons Edurne Arriola ... 17 Gadgeel S, et al. 22C3 vs

PD‐L1 immunohistochemistry as a predictive biomarker: pros and cons

Edurne ArriolaHead of Lung CancerMedical OncologyHospital del Mar

20/10/17

Page 2: PD‐L1 immunohistochemistryas a predictivebiomarker: · PDF filePD‐L1 immunohistochemistryas a predictivebiomarker: pros and cons Edurne Arriola ... 17 Gadgeel S, et al. 22C3 vs

Outline

• Current status (lung cancer)• Pros

– Predictive value for enrichment

• Cons– Lack of predictive value for exclusion– Technical issues– Complexity of immune response

Page 3: PD‐L1 immunohistochemistryas a predictivebiomarker: · PDF filePD‐L1 immunohistochemistryas a predictivebiomarker: pros and cons Edurne Arriola ... 17 Gadgeel S, et al. 22C3 vs

Currently available assays 

Hirsch F, WCLC 2017

Page 4: PD‐L1 immunohistochemistryas a predictivebiomarker: · PDF filePD‐L1 immunohistochemistryas a predictivebiomarker: pros and cons Edurne Arriola ... 17 Gadgeel S, et al. 22C3 vs

PD‐L1 scoring

Page 5: PD‐L1 immunohistochemistryas a predictivebiomarker: · PDF filePD‐L1 immunohistochemistryas a predictivebiomarker: pros and cons Edurne Arriola ... 17 Gadgeel S, et al. 22C3 vs

Are they comparable?

Blueprint 2A. M Tsao, WCLC 2017

Page 6: PD‐L1 immunohistochemistryas a predictivebiomarker: · PDF filePD‐L1 immunohistochemistryas a predictivebiomarker: pros and cons Edurne Arriola ... 17 Gadgeel S, et al. 22C3 vs

Are they comparable?Study Assays 

comparedSamples analyzed

Number of scorers

Comparabilityon TC

Comparabilityon IC

Hirsch(Blueprint 1,2)

28‐8, 22C3, SP142, SP263

39 resectedNSCLC

3 28‐8, 22C3, SP263 similar;SP142 stain fewer TC

More variable staining of IC for all assays

Scheel(german)

28‐8, 22C3, SP142, SP263

15+15 resectedNSCLC

9 28‐8, 22C3 similar; SP142 lower; SP263 higher

Low concordance(k<0.12)

Rimm(NCCN)

28‐8, 22C3, SP263

90 resected NSCLC

13 High k across 28‐8, 22C3;SP142 lower

Low correlation(k=0.19)

Ratcliffe(AstraZeneca)

28‐8, 22C3, SP263

493 excisedNSCLC

1 91‐97% agreement between assays

Not assessed

Adam (French) 28‐8, 22C3, SP142, SP263

41 resectedNSCLC

7 K>0.75 for threshold 1 and 5%

Overallagreement 75‐90%

Page 7: PD‐L1 immunohistochemistryas a predictivebiomarker: · PDF filePD‐L1 immunohistochemistryas a predictivebiomarker: pros and cons Edurne Arriola ... 17 Gadgeel S, et al. 22C3 vs

Blueprint 2A. M Tsao, WCLC 2017

22C3, 28.8 and SP263 highly concordantSP142 stains less and 73‐10 more

Tumor cells

Page 8: PD‐L1 immunohistochemistryas a predictivebiomarker: · PDF filePD‐L1 immunohistochemistryas a predictivebiomarker: pros and cons Edurne Arriola ... 17 Gadgeel S, et al. 22C3 vs

Reliability among pathologists

Moderate to strong for tumor cells

Page 9: PD‐L1 immunohistochemistryas a predictivebiomarker: · PDF filePD‐L1 immunohistochemistryas a predictivebiomarker: pros and cons Edurne Arriola ... 17 Gadgeel S, et al. 22C3 vs

What happens with IC?

Hirsch, WCLC 2017

Page 10: PD‐L1 immunohistochemistryas a predictivebiomarker: · PDF filePD‐L1 immunohistochemistryas a predictivebiomarker: pros and cons Edurne Arriola ... 17 Gadgeel S, et al. 22C3 vs

Poor reliability for immune cells

Blueprint 2A. M Tsao, WCLC 2017

Page 11: PD‐L1 immunohistochemistryas a predictivebiomarker: · PDF filePD‐L1 immunohistochemistryas a predictivebiomarker: pros and cons Edurne Arriola ... 17 Gadgeel S, et al. 22C3 vs

Do these differences have an impact in scoring?

It does for SP142 in intermediate scoring category 1‐49%

Page 12: PD‐L1 immunohistochemistryas a predictivebiomarker: · PDF filePD‐L1 immunohistochemistryas a predictivebiomarker: pros and cons Edurne Arriola ... 17 Gadgeel S, et al. 22C3 vs

Discordance is lower in high expressors

Page 13: PD‐L1 immunohistochemistryas a predictivebiomarker: · PDF filePD‐L1 immunohistochemistryas a predictivebiomarker: pros and cons Edurne Arriola ... 17 Gadgeel S, et al. 22C3 vs

Is PD‐L1 clinically relevant?

M Tsao, WCLC 2017

It is required for pembrolizumab prescription

Page 14: PD‐L1 immunohistochemistryas a predictivebiomarker: · PDF filePD‐L1 immunohistochemistryas a predictivebiomarker: pros and cons Edurne Arriola ... 17 Gadgeel S, et al. 22C3 vs

Higher expression predicts for increased benefit

Page 15: PD‐L1 immunohistochemistryas a predictivebiomarker: · PDF filePD‐L1 immunohistochemistryas a predictivebiomarker: pros and cons Edurne Arriola ... 17 Gadgeel S, et al. 22C3 vs

PembrolizumabAtezolizumab

First line therapy in PD‐L1 selected NSCLC

Carcereny WCLC 2017; Brahmer WCLC &NEJM 2017; Carbone Lancet 2017 

Nivolumab

Page 16: PD‐L1 immunohistochemistryas a predictivebiomarker: · PDF filePD‐L1 immunohistochemistryas a predictivebiomarker: pros and cons Edurne Arriola ... 17 Gadgeel S, et al. 22C3 vs

But, can we deny treatment to PD‐L1 negative (<1%)?

• In our experience 50% patients are negative (IOdetect platform)

Page 17: PD‐L1 immunohistochemistryas a predictivebiomarker: · PDF filePD‐L1 immunohistochemistryas a predictivebiomarker: pros and cons Edurne Arriola ... 17 Gadgeel S, et al. 22C3 vs

• The vast majority (77%, 115 out of 150 patients) of SP142 PD‐L1 negative patients were also PD‐L1 negative by the 22C3 assay

17 Gadgeel S, et al. 22C3 vs SP142 in OAK

SP142 Assay

Dx–, n Dx+, n

22C3 Assay

Dx–, n 115 103

Dx+, n 35 145

Dx–, no or low PD‐L1 expression; Dx+, positive for PD‐L1 expression.

N = 103 N = 115 N = 3522C3

TPS < 1%SP142

TC0 and IC0

N = 218N = 150

115 patients PD-L1–negative by 22C3 and SP142

Double negative

NegativeBy 22C3

NegativeBy SP142

Is it an antibody problem?

Page 18: PD‐L1 immunohistochemistryas a predictivebiomarker: · PDF filePD‐L1 immunohistochemistryas a predictivebiomarker: pros and cons Edurne Arriola ... 17 Gadgeel S, et al. 22C3 vs

Does not seem so…

Gadgeel et al. ESMO 2017

Page 19: PD‐L1 immunohistochemistryas a predictivebiomarker: · PDF filePD‐L1 immunohistochemistryas a predictivebiomarker: pros and cons Edurne Arriola ... 17 Gadgeel S, et al. 22C3 vs

Could it be the sampling?

Blueprint 2A. M Tsao, WCLC 2017

Page 20: PD‐L1 immunohistochemistryas a predictivebiomarker: · PDF filePD‐L1 immunohistochemistryas a predictivebiomarker: pros and cons Edurne Arriola ... 17 Gadgeel S, et al. 22C3 vs

Heterogeneity and dynamic expressionmight impact

• Expression heterogeneity: Spatial/temporal heterogeneity

Casadevall et al. CLC 2017; Uruga et al. JTO 2017; Pinato et al. Oncoimmunology 2016

Page 21: PD‐L1 immunohistochemistryas a predictivebiomarker: · PDF filePD‐L1 immunohistochemistryas a predictivebiomarker: pros and cons Edurne Arriola ... 17 Gadgeel S, et al. 22C3 vs

Other methods for PD‐L1 detectionPD‐L1 mRNA

PD‐L1 low

< median expression(n=355)

PD‐L1 high

≥ median expression(n=356)

OS HR0.68

OS HR0.74

OAK trial PD‐L1 gene expression

PD-L1 gene expression

Median OS, months Atezolizumab

Docetaxel

PD-L1 high 16.3 11.3PD-L1 low 12.7 8.7Biomarker evaluable (n=711) 14.2 9.5

Data presented in Gadgeel, et al. WCLC 2016 (Abs PL04a.02)1. Williams et al. ESMO 2016 (1171P)

2. Barlesi, et al. ESMO 2016 (Abs LBA44)

Page 22: PD‐L1 immunohistochemistryas a predictivebiomarker: · PDF filePD‐L1 immunohistochemistryas a predictivebiomarker: pros and cons Edurne Arriola ... 17 Gadgeel S, et al. 22C3 vs

Blood based PD‐L1

• PD‐L1 expression on CTCs

Page 23: PD‐L1 immunohistochemistryas a predictivebiomarker: · PDF filePD‐L1 immunohistochemistryas a predictivebiomarker: pros and cons Edurne Arriola ... 17 Gadgeel S, et al. 22C3 vs

Complexity of the immune response

Mellman Nature 2012

Page 24: PD‐L1 immunohistochemistryas a predictivebiomarker: · PDF filePD‐L1 immunohistochemistryas a predictivebiomarker: pros and cons Edurne Arriola ... 17 Gadgeel S, et al. 22C3 vs

Immune monitoring needs to be comprehensive‐immunogram

Blank, Science 2016; Yuan et al. JITC 2016

Page 25: PD‐L1 immunohistochemistryas a predictivebiomarker: · PDF filePD‐L1 immunohistochemistryas a predictivebiomarker: pros and cons Edurne Arriola ... 17 Gadgeel S, et al. 22C3 vs

Conclusions

• PD‐L1 immunohistochemical expression – When high predicts for benefit– When negative does not exclude potential benefit

• Training and validation of PD‐L1 is required for proper assessment

• Cytological samples seem adequate for testing• Heterogeneity and dynamics might play a role• It needs to be combined to other markers for optimal patient selection

Page 26: PD‐L1 immunohistochemistryas a predictivebiomarker: · PDF filePD‐L1 immunohistochemistryas a predictivebiomarker: pros and cons Edurne Arriola ... 17 Gadgeel S, et al. 22C3 vs

Gracias