PkljKYwv PDL1 Atlas 2016 PDF

INTERNATIONAL ASSOCIATION FOR THE STUDY OF LUNG CANCER
IASLC ATLAS OF PD-L1

IMMUNOHISTOCHEMISTRY
TESTING IN LUNG CANCER
EDITED BY
MING SOUND TSAO, MD, FRCPC
KEITH M. KERR, MB CHB, FRCPATH, FRCPE
SANJA DACIC, MD, PHD
Conquering Thoracic Cancers Worldwide YASUSHI YATABE, MD, PHD
FRED R. HIRSCH, MD, PHD
International Association for the Study of Lung Cancer, Aurora, CO, USA
Editors:
Ming Sound Tsao, MD, FRCPC
Keith M. Kerr, MB ChB, FRCPath, FRCPE
Sanja Dacic, MD, PhD
Yasushi Yatabe, MD, PhD
Fred R. Hirsch, MD, PhD
An IASLC publication published by Editorial Rx Press
Cover and interior design by Amy Boches, Biographics
IASLC Office:
IASLC, 13100 East Colfax Ave., Unit 10, Aurora, Colorado 80011, USA
www.iaslc.org
April 2017, October 2017

10 9 8 7 6 5 4 3 2
ISBN: 978-0-9832958-7-7
Copyright © 2017 International Association for the Study of Lung Cancer

All rights reserved
Without limiting the rights under copyright reserved above, no part of this
publication may be reproduced, stored in or introduced into a retrieval system,
or transmitted in any form, or by any means without prior written permission.
While the information in this book is believed to be true and accurate as of

the publication date, neither the IASLC nor the editors nor the publisher can
accept any legal responsibility for any errors or omissions that may be made.
The publisher makes no warranty, express or implied, with response to the
material contained therein.
Erratum: The images in Figures 8 and 10 in Chapter 3, “Immunohistochemistry

for PD-L1” (page 47) were reversed in the original publication of this PD-L1 Atlas.
These images are correct in this printing of the publication.
EDITED BY
MING SOUND TSAO, MD, FRCPC
KEITH M. KERR, MB CHB, FRCPATH, FRCPE
SANJA DACIC, MD, PHD
YASUSHI YATABE, MD, PHD
FRED R. HIRSCH, MD, PHD
AN INTERNATIONAL ASSOCIATION FOR THE STUDY OF LUNG CANCER PUBLICATION
Editorial Rx Press
North Fort Myers, FL
Acknowledgments
IASLC acknowledges the generous funding and support

provided by AstraZeneca, Bristol-Myers Squibb, and Merck
for the IASLC Atlas of PD-L1 Immunohistochemistry Testing in
Lung Cancer.
The coeditors and contributors also acknowledge the
assistance of Dr. Murry Wynes, PhD, Scientific Affairs Director,
IASLC, for coordinating the project; the editorial assistance of
Joy Curzio and Lori Alexander, MTPW, ELS, MWC; and, the
publishing support of Deb Whippen, Editor and Publisher,
Editorial Rx Press, for the publication of this text.
Contents
Contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Manufacturers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1 Tumor Immunology. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2 Cancer Immunotherapy for Lung Cancer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .21
3 Immunohistochemistry for PD-L1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
4 PD-L1 28-8 pharmDx Assay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .49
5 PD-L1 22C3 pharmDx Assay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .55
6 PD-L1 SP142 Assay. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .63
7 PD-L1 SP263 Assay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .73
8 PD-L1 73-10 Assay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79

9 Other Anti–PD-L1 Clones: Alternative Assays and Laboratory-Developed Tests . . . . . . . . . . . . 83
10 Complementary and Companion Diagnostics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
11 Assay Harmonization: Is It Possible? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .97
12 Implementation of PD-L1 Testing for Personalized Therapy for Lung Cancer . . . . . . . . . . . . . . 109
13 Summary and Future Perspectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
6 IASLC ATLAS OF EGFR TESTING IN LUNG CANCER
Contributors
Editors Contributing Authors

Ming Sound Tsao, MD, FRCPC Bernadette Reyna Asuncion, MD
Pathologist, Senior Scientist, and Professor Research Fellow
M. Qasim Choksi Chair in Lung Cancer Cancer Science Institute of Singapore
Translational Research National University of Singapore
Princess Margaret Cancer Centre, University Singapore, Singapore
Health Network
Department of Laboratory Medicine and Elisabeth Brambilla, MD, PhD
Pathobiology Professor
University of Toronto Department of Pathology
Toronto, Canada Centre Hospitalier Universitaire de Grenoble
Grenoble, France
Keith M. Kerr, FRCPath
Professor Lukas Bubendorf, MD
Department of Pathology Professor and Head
Aberdeen University Medical School, Division of Cytopathology
Aberdeen Royal Infirmary Institute for Pathology
Aberdeen, Scotland, United Kingdom University Hospital Basel
Basel, Switzerland
Sanja Dacic, MD, PhD
Professor Reinhard Buettner, MD
Director of the FISH and Developmental Professor and Chairman
Laboratory Institute for Pathology
Department of Pathology University Hospital Medical School
University of Pittsburgh Medical Center University of Cologne
Pennsylvania, U.S.A. Cologne, Germany
Yasushi Yatabe, MD, PhD Teh-Ying Chou, MD, PhD

Chief Professor
Department of Pathology and Molecular Institute of Clinical Medicine
Diagnostics National Yang-Ming University
Aichi Cancer Center Department of Pathology and Laboratory
Nagoya, Japan Medicine
Taipei Veterans General Hospital
Fred R. Hirsch, MD, PhD Taipei, Taiwan
Professor
Department of Medicine Wendy A. Cooper, MBBS, Bsc(Med), FRCPA, PhD
Department of Pathology Associate Professor
Pia and Fred R. Hirsch Endowed Chair Tissue Pathology and Diagnostic Oncology,
University of Colorado at Denver Royal Prince Alfred Hospital
CEO, International Association for the Study Sydney Medical School, University of Sydney
of Lung Cancer (IASLC) School of Medicine, Western Sydney University
Colorado, U.S.A. Sydney, Australia
CONTRIBUTORS 7
Sylvie Lantuéjoul, MD, PhD Ross A. Soo, MBBS

Department of Pathology Principal Investigator
Pôle de Biologie, Institut de Biologie et de Cancer Science Institute of Singapore
Pathologie National University of Singapore
CHU A Michallon, CS 10217 Senior Consultant
Université Joseph Fourier- INSERM U 823 National University Health System
Institut Albert Bonniot Singapore
Grenoble, France
Erik Thunnissen, MD, PhD
Mari Mino-Kenudson, MD Consultant Pathologist
Director, Pulmonary Pathology Service VU University Medical center
Massachusetts General Hospital Amsterdam, The Netherlands
Associate Professor of Pathology
Harvard Medical School Arne Warth, MD, PhD
Massachusetts, U.S.A. Institute of Pathology
Heidelberg University Hospital
Andrew G. Nicholson, DM Heidelberg, Germany
Professor
Department of Histopathology Ignacio Wistuba, MD
Royal Brompton and Harefield NHS Professor and Chair
Foundation Trust and Imperial College Jay and Lori Eisenberg Endowed Professor
London, United Kingdom Department of Translational Molecular
Pathology
Lynette M. Sholl, MD The University of Texas MD Anderson
Assistant Professor Cancer Center
Department of Pathology Texas, U.S.A.
Brigham and Women’s Hospital
Harvard Medical School Murry W. Wynes, PhD
Massachusetts, U.S.A. Director of Scientific Affairs
International Association for the Study of
Lung Cancer
Colorado, U.S.A.
Left to right: Teh-Ying Chou, Yasushi Yatabe, Lukas

Bubendorf, Elisabeth Brambilla, Ming Sound Tsao, Mari
Mino-Kenudson, Sylvie Lantuéjoul, Erik Thunnissen,
Sanja Dacic, Keith M. Kerr, Pia Hirsch, Ross A. Soo, Fred
R. Hirsch, and Murry Wynes. Not present: Bernadette
Reyna Asuncion, Reinhard Buettner, Wendy A. Cooper,
Andrew G. Nicholson, Lynette M. Sholl, Arne Warth, and
Ignacio Wistuba.
Abbreviations
The following abbreviations are used in the text.
AEC: 3-amino-9-ethylcarbazol
ALK: anaplastic lymphoma kinase (gene)
APC: antigen presenting cell
CTLA-4: cytotoxic T-lymphocyte-associated antigen 4
DAB: 3, 3’ diaminobenzidine
DC: dendritic cell
EDTA: ethylenediaminetetraacetic acid
EGFR: epidermal growth factor receptor (gene)
ER: estrogen receptor
FDA: US Food and Drug Administration
FFPE: formalin-fixed paraffin-embedded
H&E: hematoxylin & eosin
HER2: human epidermal growth factor receptor-2 (gene)
HLA: human leukocyte antigen
HRP: horseradish peroxidase
ICC: immunocytochemistry
IFN-g: interferon gamma
IgG: immunoglobulin G
IHC: immunohistochemistry
IL-2: interleukin-2
KIF5B: kinesin family member 5B (gene)
KIR: killer cell immunoglobulin-like receptor
LAG3: lymphocyte-activation gene 3
LDT: laboratory-developed test
MAGE-A3: melanoma-associated antigen-A3
MHC: major histocompatibility complex
MUC1: mucinous glycoprotein-1
NSCLC: non-small cell lung cancer
PD-1: programmed cell death protein-1
PD-L1: programmed cell death ligand-1
PR: progesterone receptor
ROS1: c-ros oncogene 1 (gene)
SCLC: small cell lung cancer
TGF-b2: transforming growth factor beta 2
TIL: tumor infiltrating lymphocyte
TNF-α: tumor-necrosis factor alpha
MANUFACTURERS 9
Manufacturers
The following manufacturers and their PD-L1 testing-related products are noted in this
Atlas. The location given for each manufacturer is not the only location; most manufacturers
have offices worldwide.
Abcam Leica Biosystems

Cambridge, UK Buffalo Grove, Illinois, USA
Antibody clone 28-8, rabbit polyclonal anti PD-L1 BOND-MAX Automated IHC/ISH Stainer;
ab58810 Novolink Polymer Detection System
Agilent Technologies/Dako Proteintech Group, Inc.

Carpineteria, California, USA Rosemont, Illinois, USA
PD-L1 IHC 22C3 pharmDx; PD-L1 IHC 28-8 PD-L1 rabbit polyclonal CD274 antibody
pharmDx; Autostainer Link 48; PT Link
Pre-Treatment Module; EnVision FLEX Target R&D Systems Inc.
Retrieval Solution; EnVision FLEX+ Polymer Minneapolis, Minnesota, USA
Reagents; EnVision FLEX+ Wash Buffer; Mouse monoclonal MAB1561
FLEX IHC microscope slides
Sigma-Aldrich
Anatech LTD St. Louis, Missouri, USA
Battle Creek, Michigan, USA Anti-CD274 rabbit antibody
Prefer fixative
Thermo Fisher Scientific
BioCare Medical Waltham, Massachusetts, USA
Pacheco, California, USA Fisherbrand Superfrost Plus Microscope Slides;
DaVinci Green Diluent mouse monoclonal MIH1; Tris-EDTA buffer
solution; UltraVision Quanto Detection System
Bio SB HRP DAB
Santa Barbara, California, USA
PD-L1/CD274 clone: RBT-PDL1 rabbit Ventana Medical Systems, Inc.
monoclonal Tucson, Arizona, USA
PD-L1 (SP263) Assay; PD-L1 (SP142) Assay;
Cell Signaling Technology, Inc. Benchmark ULTRA platform, OptiView DAB IHC
Danvers, Massachusetts, USA Detection Kit, OptiView Amplification Kit, Rabbit
E1L3N rabbit monoclonal antibody; SignalStain Monoclonal Negative Control Ig; ChromoMAP
Boost IHC Detection Reagent DAB; Benchmark XT Autostainer; Cell
Conditioning 1 (CC1); ultraView Universal DAB
Enzo Life Sciences Inc. Detection Kit; DISCOVERY ChromoMap DAB Kit
Farmingdale, New York, USA
EDTA pH 8
Introduction
By Ming S. Tsao, Keith M. Kerr, Sanja Dacic, Yasushi Yatabe, and Fred R. Hirsch
Despite very encouraging progress in the development and use of immunotherapy for
patients with non-small cell lung cancer, much confusion remains regarding patient selection
for each therapy. Programmed cell death ligand-1 (PD-L1) protein expression, as detected
by immunohistochemistry (IHC) testing, has been widely used as a predictive biomarker
assay for anti–PD-1/PD-L1 therapies. In fact, an assay for determination of PD-L1 expres-
sion is approved by the US Food and Drug Administration for both first-line and second-line
therapy with pembrolizumab. There is no clear understanding among physicians, health
care personnel, or patients, however, regarding which assay to use for PD-L1 testing and
whether the various assays are interchangeable because each assay was co-developed with
a therapy. This complex biomarker scenario—the likes of which we have not faced before in
lung cancer diagnostics—poses many challenges for pathologists, oncologists, and patients.
The International Association for the Study of Lung Cancer (IASLC) has recognized the
importance and timeliness of this topic and has convened an expert panel of authors to pres-
ent current information about the emerging PD-L1 IHC assays, as well as to highlight both
areas of clarity and debate. The authors have approached this topic with a wider lens, looking
at the changing landscape of laboratory testing in general, as well as with a detailed focus
on the specifics of each assay and on the current controversies regarding PD-L1 expression
testing in lung cancer. Although this Atlas primarily aims to be a guide or resource for phy-
sicians and others involved in lung cancer diagnosis and treatment, it is our hope that this
text eventually also may give patients a more comprehensive understanding of the current
biomarker scenario. Ultimately, we hope that through the creation of this Atlas, patients
with lung cancer will receive the most contemporary and well-suited treatment options,
based on up-to-date evidence, and will feel more confident and knowledgeable regarding
their therapy.
The authors acknowledge that updates to this Atlas will almost certainly be needed,
sooner rather than later, due to the rapidly evolving nature of the field. Other biomarkers
relating to the immune response itself or to tumor mutational burden are being investigated.
Whether these will prove to be superior to PD-L1 IHC testing as a guide for therapeutic
selection remains to be seen. In the meantime, PD-L1 IHC is the validated biomarker of
choice. There are numerous ongoing trials investigating this biomarker and its associated
analytic tools, and it seems likely that it will be at least part of the biomarker profile required
for administration of anti–PD-1/PD-L1 drugs for the foreseeable future.
Tumor Immunology
By Yasushi Yatabe, Elisabeth Brambilla, and Keith M. Kerr 1
It is widely accepted that cancer develops because of the accumulation of various alterations,
including genetic and epigenetic changes, that make cancer cells genotypically and pheno-
typically different from normal cells. One such example is the expression of cancer-testis
antigens in several solid tumors, including lung cancer (Rousseaux 2013). These antigens are
normally expressed in early embryonic and germ cells but silenced in adult somatic cells.
Disrupted DNA methylation patterns of promoter CpG islands in cancer cells lead to aberrant
expression; thus, expression of the cancer-testis antigens is restricted to cancer cells. More
than 100 gene families with such an expression pattern have been identified. The antigens
can be recognized by the host immune system and induce an immune response, although
the testis is protected from immune attack by a lack of major histocompatibility complex
(MHC) class I and II molecule expression. Furthermore, gene mutations and amplification
may change the protein structure and expression level, which are also capable of being immu-
nogenic. The number of genetic mutations in a tumor, the mutation burden, is associated
with neoantigen burden (Rizvi 2015). The host immune system recognizes and responds to
these antigens to a certain extent. However, cancer cells can find ways to survive through
the acquisition of tolerance mechanisms, thus escaping from immune recognition. Under
the current hypothesis, the immune system initially recognizes cancer cells and induces
an immune response. After the equilibrium between cancer cell elimination by so-called
immune surveillance and cancer cell evolution by genetic instability, the tumor cell clone
is either eliminated or cancer cells survive but remain dormant. This dormancy is due to a
decreased immunogenic state with adaptation within the cancer microenvironment, known
as immunoediting (Schreiber 2011). Immune escape must then occur for a clinically evident
tumor to develop. In the following sections, mechanisms of escape from immune surveil-
lance in cancer are discussed, with reference to immunotherapeutic approaches (Box 1).
Cancer and T-Cells

Even within the same species, organ transplantation causes immune responses. This
fact implies that the immune system distinguishes self-antigens from non-self-antigens.
14 IASLC ATLAS OF PD-L1 IMMUNOHISTOCHEMISTRY TESTING IN LUNG CANCER
T cells recognize an antigen, which is Box 1. Major Immunotherapeutic Approaches

presented with MHC by antigen-pre-
Adaptive immunotherapy
senting cells (APCs), such as dendritic • Passive transfer of the immune cells with anticancer
cells (DCs; Figure 1). Phagocytosed activity, such as tumor-associated antigen (TAA)-specific
antigens are processed to the peptides T-cell clones and tumor-infiltrating lymphocytes
and presented onto MHC in the sur- • Genetically engineered immune cells
Cancer vaccine
face of the APCs. As a result of this
• Immunization to enhance antitumor reactions
presentation, T cells that have T-cell
Nonspecific stimulation of immune responses
receptors specific to the antigens rec- • Stimulation of effector cells
ognize the antigens and are activated • Inhibition of regulatory cells
in coordination with costimulatory
receptors (CD28 and 4-1BB). In the early studies of tumor challenge after immunization in
mice, CD8+ cytotoxic T-cells were a major player in mediation of tumor rejection. Using this
function of T cells, researchers have attempted adaptive T-cell immunotherapy. The therapy
is based on ex vivo expansion of patient-derived tumor-specific T cells and reinfusion to
the patients. Peripheral mononuclear cells in the blood are isolated and stimulated with DCs
that have been exposed to peptides of tumor-associated antigens. Through repeated stimu-
lation and expansion, specific T-cell clones are collected and reinfused into the patients.
This approach has some advantages related to cancer-specific activity and irrelevant immu-
nosuppressive tumor microenvironment. Similarly, tumor infiltrating lymphocytes (TILs)
are used with adaptive immunotherapy because lymphocytes with anticancer activity are
likely to be enriched. It has been reported that complete remission was achieved with this
method in 20 of 93 patients with metastatic melanoma, and 19 patients have experienced
ongoing complete regression for 3 years (Rosenberg 2011). In this study, TILs were collected
from resected melanoma, and T-cell clones with optimal anticancer activity were isolated,
DC MHC Class I
TCR CTL
Peptide
CD8
MHC Class II
Peptide
TCR
s
Cytokine CD4
B cell
Cancer
Antibodies
Figure 1. Immune reaction against cancer cells is initiated by interaction between T-cell receptors (TCR) and major histo-
compatibility complex (MHC) molecules, the latter presenting processed peptides of immunogens. DC = dendritic cells,
and CTL = cytotoxic T-lymphocytes.
TUMOR IMMUNOLOGY 15
followed by expansion with interleukin-2 (IL-2) stimulation and reinfusion immediately

after lymphodepleting chemotherapy using cyclophosphamide and fludarabine, with or
without total body radiation. Because obtaining TILs with anticancer activity tends to be
difficult in cancers other than melanoma, genetically engineered immune cells that specifi-
cally recognize cancer cells have also been examined, and the reinfusion of the lymphocytes
with exogenous high affinity to cancer cells has been shown to achieve objective clinical
responses (Morgan 2006).
Dendritic Cells
As with many vaccines, it would be expected that active immunization against cancer-spe-
cific antigens would provoke cellular immune recognition to inhibit the growth of established
cancer. DCs play a key role in the induction of T-cell responses through presentation of the
target peptides on MHC molecules (Figure 1). In DCs, phagocytosed antigens are processed
into peptides by the proteasome in the cytosol. The complexes are moved via the endoplas-
mic reticulum through special channels, and the processed peptides are loaded onto MHC
molecules. Lastly, the MHC molecules that present the peptides are expressed on the cell
surface. Therefore, it should be efficient to use DCs for cancer vaccination.
A common method for generating the vaccine is as follows. DC precursors are obtained
from bone marrow or peripheral blood mononuclear cells and are differentiated into imma-
ture DCs with stimulation of granulocyte macrophage colony-stimulating factor (GM-CSF)
and/or interleukin-4 (IL-4), followed by exposure to peptides to generate mature DCs.
Several methods, including fusion of the DCs with tumor cells, and co-stimulation with toll-
like receptor (TLR) ligands and/or agonistic anti-CD40 antibody may be used to enhance
the maturation.
Sipuleucel-T was approved as a vaccine therapy for patients with castration-resistant
prostatic cancer by the US Food and Drug Administration (FDA) based on the results of a
phase III clinical trial (Kantoff 2010). With this treatment, peripheral blood mononuclear
cells are collected and incubated with a fusion protein of prostatic acid phosphatase and
GM-CSF, as a cancer-associated antigen and an antigen-presentation activator, respectively.
Antigen-pulsed APCs were reinfused once a week for one month. With this vaccine therapy,
the relative risk of death was reduced by 22%, although the time to disease progression was
similar for the treatment and placebo arms. However, Sipuleucel-T is exceptional, as most
cancer vaccine therapies have failed to show clinical effectiveness. In a summary of findings
for cancer vaccine trials of 440 patients in the National Cancer Institute Center for Cancer
Research Surgery Branch, the authors report that the objective response rate was 2.6%, and
similar results were observed in other studies (Rosenberg 2004).
Coordination of Immune Responses

A remarkable characteristic of the immune system is its ability to recognize and eliminate
the targets specifically. The features are mediated by complex mechanisms involving T cells,
DCs, and other immune cells by a balance between co-stimulatory and inhibitory signals
(Figure 2). Recent developments in the understanding of cancer immunology allow the use of
such immune coordination mechanisms for cancer management by means of enhancement
of T-cell responses or by blockage of the negative regulation of T-cell responses.
Cytokines Stimulating Antigen-presenting cell T cell

Effector Cells PD-1 –
PD-L1 or PD-L2
As described previously, cytotoxic
T cells developed from naïve CD8+ CD80 or CD86 CD28 +
T cells have the ability to directly CD80 or CD86 CTLA-4 –

eliminate the targets that express
peptides with MHC class I mol- B7RP1 ICOS +
ecules. In contrast, CD4 T cells

+
recognize peptides presented by HVEM BTLA –
MHC class II molecules and mediate KIR –

Peptide
T-helper cell functions through dif- Signal 1
MHC class I or II TCR
ferentiation of helper T cells to the
LAG3 –
distinctive T-helper type subsets.
CD137L CD137 +
These subsets include T-helper type
1 (Th1) for enhancement of cyto- OX40L OX40 +
toxic T-lymphocyte (CTL) function,
CD70 CD27 +
T-helper type 2 (Th2) for B-cell
responses, T-helper type 17 (Th17) CD40 CD40L
for autoimmunity and tissue inflam-
mation, and regulatory T cells (Treg GAL9 TIM3 –
cells) for suppression of immunity. Adenosine

A2aR –
The differentiation of T cells and
the subsequent immune responses
require interaction with various Figure 2. T-cell responses are affected by multiple immune modula-
cytokines. Th1 cells produce high tors. Most co-stimulatory receptors are expressed on naive and resting
T cells, whereas co-inhibitory receptors are commonly upregulated after
levels of the cytokines IL-2, tumor T-cell activation. Modified, with permission from Pardoll DM. The block-
necrosis factor-alpha (TNF-α) and ade of immune checkpoints in cancer immunotherapy. Nat Rev Cancer.
interferon gamma (IFN-γ), whereas 2012;12(4):252-264.
interaction of Th2 helper cells with
IL-4 results in STAT6 activation followed by activation of GATA-3, which is essential for
Th2 helper functions. Although individual cytokines target specific functions and particu-
lar effectors, IL-2 stimulates a wide range of T cells, and it was discovered to be a factor
that primarily stimulates proliferation of T cells (Morgan 1976). Because T-cell responses
play a major role in antitumor activity, exogenous administration of IL-2 was expected to
mediate tumor regression via activated T-cell responses. Although this immunotherapy
was effective in a murine model, no tumor regression was shown in an early study of 20
patients with cancer who were treated with recombinant IL-2 (Lotze 1985). The pharma-
cokinetics of the study indicated depletion of all lymphoid cells just after administration of
IL-2, suggesting stimulation of Treg cells that dampened effector T-cell responses against
tumor antigens. However, subsequent treatment with high-dose IL-2 achieved the first
marked clinical responses to immunotherapy (Rosenberg 1985). The FDA approved the
treatment based on the findings of several phase II trials involving patients with metastatic
renal cell carcinoma that showed a 14% overall response rate (9% partial and 5% complete
responses (Fyfe 1995). Of note, many responses with this therapy lasted for more than 5
TUMOR IMMUNOLOGY 17
years, suggesting remarkable durability of anticancer responses in contrast to other cyto-

toxic chemotherapies. In addition to being approved for metastatic renal cell carcinoma,
this IL-2 therapy was approved for advanced melanoma; a 16% overall response rate and
flat tails in the Kaplan-Meier curve were also reported (Atkins 1999). Major toxicities of
this therapy are due to the responses mediated by IL-2-induced IFN-γ and TNF-α, which
result in capillary leak syndrome and decreased systemic vascular resistance. These, in turn,
lead to fever, hypotension, arrhythmia, lethargy, renal failure, and systemic edema. Other
immunotherapy, through stimulation of effector cells, includes treatment using IFN-α and
imiquimod; however, this treatment resulted in clinical efficacy only for some cancer types
(Kirkwood 1996, Motzer 2002, van Seters 2008).
Box 2. Mechanisms of Escape from Immune Surveillance

Immune Checkpoint Inhibitors
Cancer cells can survive even in Inhibition of regulatory T-cells (Treg cells)
It has been demonstrated in many studies that tumor-derived
immunocompetent individuals
Treg cells have comparatively higher suppressive activity than
because the cells acquire toler- naturally occurring Treg cells.
ance mechanisms that allow them
Defective antigen presentation
to escape immune surveillance, Cytotoxic T lymphocytes cannot recognize target antigens on
with various mechanisms being cancer cells by impaired MHC I pathway, proteasome subunits
proposed (Box 2). Although early LMP2 and LMP7, TAP, and tapasin.
attempts using the immune system Immune suppressive mediators
were mostly focused on boost- Cancer cells and/or the microenvironment altered by cancer
cells produce immunosuppressive cytokines including TGF-beta,
ing immune attack, recent results TNF-a, IL-1, IL-6, CDF1, IL8, and IL-10.
have demonstrated that so-called
Dysregulation of co-stimulatory and co-inhibitory molecules
releasing the brakes—including Cancer cells downregulate co-stimulatory molecules, such
inhibition of immune check- as CD28, and induce expression of co-inhibitory molecules, such
points—is effective against cancer. as PD-L1.
Two major pathways, cytotoxic Abbreviations: MHC, major histocompatibility complex; TAP, transporter associated
T-lymphocyte–associated antigen with antigen processing; TGF-beta, transforming growth factor beta; TNF-α , tumor
necrosis factor-alpha; IL, interleukin; PD-L1, programmed death-ligand 1.
4 (CTLA-4) and programmed cell
death protein-1 (PD-1), have received much attention because of remarkable efficacy in
numerous clinical trials for various cancer types (Pardoll 2012, Ott 2013).
CTLA-4
An anti–CTLA-4 was the first immune checkpoint inhibitor approved by the FDA, based
on the results of the phase III clinical trials in which the CTLA-4 antagonist ipilimumab
improved overall survival for patients with previously untreated metastatic melanoma
(Robert 2011, Hodi 2012). CTLA-4 is expressed exclusively on T-cells, where it primarily
down-modulates the amplitude of T-cell activation. As previously described, antigen recog-
nition initiates T-cell activation through engagement of MHC-bound antigens on APCs with
the T-cell receptor, followed by co-stimulation via CD80/CD86-CD28 interactions. In paral-
lel, inhibitory signals mediated by CTLA-4 dampens the reaction by outcompeting CD28
for binding of CD80/CD86 molecules, inhibiting IL-2 production and preventing cell-cycle
progression. Because stronger antigen stimulation through T-cell receptors leads to greater
amounts of CTLA-4 expression, this inhibitory system functions as a signal dampener to
maintain a consistent level of T-cell activation (Figure 3). In fact, massive lymphoprolif-
eration and systemic immune hyperactivation have occurred in CTLA-4 knockout mice
(Tivol 1995, Waterhouse 1995). Detailed mechanisms of CTLA-4 blockage have not been
elucidated, but it is clear that tumors can use this pathway to evade immune surveillance,
as highlighted by clinical benefit of CTLA-4 inhibition in some tumor types.
PD-1 and PD-L1

Both programmed cell death ligand-1 (PD-L1) and PD-L2 are members of the B7 family
and bind to PD-1. The two molecules share 37% sequence homology and arose through
gene duplication, which has positioned them within 100 kb of each other in the genome. In
contrast to predominant expression of PD-L2 on APCs, PD-L1 can be expressed in various
cells including T cells, epithelial cells, and endothelial cells. Although CTLA-4 works in
the initial phase of T-cell recognition, the PD-L1 pathway plays a role in the latter phase of
the immune response, such as within inflammatory tissues, to regulate T-cell function and
prevent autoimmunity. In the case of microorganism infection, the foreign antigens activate
T cells, which in turn upregulate PD-1 expression on the T-cell surface. Also, inflammatory
signals in the tissues induce the PD-L1 expression, which prevents collateral tissue damage
via T-cell inhibition (Figure 3). It is also known that excessive PD-1 expression, which is
typically induced by chronic antigen exposure, is associated with an exhausted or anergic
state in T cells. In cancer tissues, PD-1 is upregulated on TILs, while the ligand, PD-L1, is
expressed on many cancer cell types. PD-L1 expression is often associated with a poor
outcome (Sznol 2013). There appears to be an intrinsic adaptive response, in that cancer
A
CD80 Strong stimulation leads
or CD86 CD28 to CTLA-4 expression
+
DC +
–
TCR
Peptide CTLA-4
MHC Naive or resting T cell CTLA-4 dampens
(CD28+, CTLA-4) the stimulation
Co-stimulating
B receptor
Trafficking
Activated T-cells
upregulate PD-1
Co-stimulating
ligand of T cells to
peripheral Tissue
tissues or
cancer
+ cells
–
PD-L1 PD-1
Priming or PD-L2
of T cells
Chronic antigen
Inflammatory signals, such exposure can induce an
as IFN-γ induce PD-L1 anergic state in T cells
Figure 3. Two major immune checkpoint regulators and the immune responses. A. Cytotoxic T lymphocyte-associated antigen
4 (CTLA-4). B. Programmed cell death protein 1 (PD-1). Modified with permission from T-cell responses are affected by multiple
immune modulators. Most co-stimulatory receptors are expressed on naive and resting T cells, whereas co-inhibitory recep-
tors are commonly upregulated after T-cell activation. Modified with permission from Pardoll DM. The blockade of immune
checkpoints in cancer immunotherapy. Nat Rev Cancer. 2012;12(4):252-264. DC = dendritic cells, MHC = major histocompat-
ibility complex, TCR = T-cell receptor, PD-L1 = programmed cell death ligand 1, PD-L2 = programmed cell death ligand 2.
TUMOR IMMUNOLOGY 19
cells express PD-L1 to escape from immune

surveillance via ligation of PD-1-expressed Cancer
T cell
TILs (Figure 4). However, the results of some Peptide
MHC
TCR
studies suggest that constitutive oncogenic sig-
nals promote PD-L1 expression in cancer cells.
ALK gene rearrangement, which was initially Effector function
discovered in lymphoma, was shown to induce
PD-L1 expression through the STAT3 pathway
(Marzec 2008). Regardless of the mechanisms of Elimination by
Immune surveillance
PD-L1 induction, PD-1 expression on TILs and
increased expression of PD-L1 on cancer cells
provide a reasonable rationale for targeted ther- Cancer
T cell
apy of these molecules. As discussed in Chapter Peptide
MHC
TCR
2, immunotherapy-targeted to PD-1 or PD-L1 has
been led to significant clinical efficacy in various
cancer types, including lung cancer. PD-1 PD-L1
Inhibition
Mechanism of Immune Escape in Lung Cancer
Although lung cancer is one of the most
Effector function
molecularly complex cancers (second only to
-melanoma), many of these genetic changes Figure 4. Programmed cell death protein 1 (PD-1)/
programmed cell death ligand 1 (PD-L1) pathway and
may not elicit functional immune recognition cancer. T cells attack cancer cells through the effector
and attack by the CD8+/PD-1+ cytotoxic cells function (A). However, cancer cells can escape from
for several reasons. immune surveillance with expression of PD-L1. Activated
T-cells express PD-1. Ligation of PD-1 with PD-L1 down-
regulates the effector function (B). TCR = T-cell receptors,
Insufficient load of neoantigens that engage T-cell MHC = major histocompatibility complex.
/MHC Class I-specific recognition.
Across various cancer types, cytolytic activity of immune cells was highest in renal clear
cell carcinoma and cervical cancer when the activity was elucidated by tissue expression of
granzyme A and perforin using TCGA data. Lung cancer was ranked in the middle, accord-
ing to this mechanism (Rooney 2015).
Downregulation or complete loss of MHC complex molecule expression on tumor cells to bind
the T-cell receptor.
Downregulation or loss of MHC complex molecules, including human leukocyte antigen
(HLA) class I, β2-macroglobulin, and antigen-processing machinery components, has been
reported among various cancer types and was seen in 60% to 94% of patients with non-small
cell lung cancer (NSCLC). Loss of this expression was associated with TILs, particularly
CD8+ T cells (Garrido 1995, Kikuchi 2007, Ramnath 2006, Garcia-Lora 2003).
Low density of CD8+/PD-1+ active CTLs in the immune infiltrate, or a lack of immune infiltration.
It has been known that some soluble factors, which are produced by tumor cells, suppress
local immune responses. The capacity for antigen presentation was reduced in IL-10-
deficient mice (Hagenbaugh 1997). Actually, dense infiltration of lymphocytes was rare in
lung cancer (Brambilla 2016), and low density of active CD8+/PD-1+ CTLs has been reported
(Tumeh 2014, Kim 2015).
Anergy of CD8 cytotoxicity.

In contrast to some reports of low density of CD8+/PD-1+ cytotoxic T cells, some tumors
have high numbers of infiltrating CD8+ cytotoxic T cells. This finding is explained by anergy,
which results from T-cell inactivation, with a lack of granzyme B, a lack of proper cytokines
for achieving CD8+ T-cell maturation and activation (IFN- γ, IL-2, IL-21), or the presence of
inhibitory cytokines, such as IL-10 and transforming growth factor beta (Zaretsky 2015).
Counter-regulation by excessive CD25+FOXP3+CD4+ on CD8+PD-1+ Treg cells in the immune

infiltrate.
A high ratio of intratumoral Tregs to effector T cells is generally associated with poor out-
comes across many cancer types, including lung cancer (Fridman 2012, Petersen 2006).
Unavailability of functional apoptotic pathways in tumor cells by lack of functional Fas-FasL

receptor- complex.
Approximately 20% to 25% of lung cancers lack Fas expression and overexpress FasL, sug-
gesting impaired cytokilling via the Fas/FasL pathway (Li 2015, Viard-Leveugle 2003).
Immune checkpoints with increased PD-L1 or CTLA-4 on tumor cells and or immune cells.
As discussed previously, expression of PD-L1 and CTLA-4 on cancer cells allows immune
cells to fail to react to tumor-cell antigens.
Summary
Even from this brief and short review, it is clear that the interactions of tumor cells and the
immune system are extremely complex and not entirely understood. These interactions
have a pivotal role in allowing tumors to develop and progress. Primarily, tumors, by one
or several mechanisms, must develop the ability to avoid or negate an immune response in
those cases in which a specific immune response to tumor neoantigens has been developed.
Among those mechanisms are the interaction of membrane-bound ligands and receptors
that act as immune checkpoints, regulating the immune system. This important physiologic
mechanism—which prevents uncontrolled immune responses and, thus, autoimmunity—
appears to be adopted by some tumors as a means of switching off an otherwise primed and
available cellular immune response to that tumor. These receptor–ligand interactions are,
therefore, important therapeutic targets, as evidenced by the successes seen with the use
of anti-CTLA-4 therapies in melanoma and anti-PD-1 or PD-L1 agents in a number of tumor
types, such as NSCLC. In some ways, it is remarkable, in such a complex, multifaceted, and
closely regulated system, that inhibiting only one regulatory mechanism can achieve such
results. This atlas will discuss the inhibition of the PD-1/PD-L1 axis in lung cancer, the
clinical evidence to date for such therapies, and the challenges posed by a very complex
biomarker backdrop to this exciting new therapeutic approach.
Cancer Immunotherapy for Lung Cancer
By Ross A. Soo, Murry W. Wynes, and Fred R. Hirsch 2
Lung cancer is the leading cause of cancer death worldwide, with 1.6 million attributed
deaths annually (World Health Organization International Agency for Research on Cancer,
2017). Non-small cell lung cancer (NSCLC) accounts for the majority of lung cancer diag-
noses, and the disease is metastatic at the time of diagnosis for most patients (National
Cancer Institute Surveillance, Epidemiology, and End Results Program, 2017). Despite an
improvement in overall survival with platinum-based chemotherapy (NSCLC Meta-Analyses
Collaborative Group, 2008), prognosis remains poor for patients with advanced-stage NSCLC,
with a median survival of 8 to 12 months (Schiller 2002, Sandler 2006). Advances in the
molecular characterization of NSCLC, especially in adenocarcinoma histologic subtypes,
have enabled the identification of key genetic aberrations in NSCLC that can be exploited
with molecularly targeted therapy (Pao 2011). Genetic aberrations in EGFR, ALK, ROS1,
RET, BRAF, and NTRK predict for sensitivity to receptor tyrosine-kinase inhibitors (Mok
2009, Solomon 2014, Shaw 2014, Planchard 2016). Despite the success of molecularly targeted
treatment, acquired resistance and disease progression inevitably occur (Camidge 2014;
Hirsch 2016). Treatment options for patients with small cell lung cancer (SCLC) in whom
disease has progressed after platinum-based chemotherapy are even more limited. Novel
therapeutic approaches are needed for patients with NSCLC and SCLC.
Cancer immunotherapy has been described as any therapy that interacts with the immune
system to treat cancer. As an option for cancer, cancer immunotherapy predates even cyto-
toxic chemotherapy. Cancer immunotherapy can be categorized into passive and active types
(Figure 1). Passive immunotherapy has been described as administration of an immunologi-
cally active agent manufactured or generated outside of the patient’s body. Theoretically,
such an approach is not dependent on the host’s own immune system to have an effect.
Examples of passive immunotherapy include the use of monoclonal antibodies, such as
trastuzumab or rituximab (Slamon 2001, Coiffier 2002), and adoptive cellular therapy, such
as tumor-infiltrating lymphocyte infusion, T-cell receptor (TCR) engineering, and chimeric
antigen receptor T-cell therapy (Morgan 2006, Maude 2014). Active cancer immunotherapy
involves the stimulation or priming of the host’s immune system to recognize a tumor as
foreign. Examples of active immunotherapy include cancer vaccination with tumor antigens
and an adjuvant enhancement of immune cell function with cytokines, as well as targeting
of immune checkpoint regulatory receptors with immune checkpoint inhibitors (Figure 1).
This chapter provides an overview of the role of cancer vaccines and the check point
inhibitors that target cytotoxic T lymphocyte-associated protein 4 (CTLA-4), and pro-
grammed cell death protein-1 (PD-1)/programmed cell death ligand-1 (PD-L1) in NSCLC
and SCLC. Studies examining the efficacy of cytokines, such as interferon alpha and inter-
leukin-2 (IL-2), in patients with lung cancer have been negative and will not be discussed
(Jansen 1992, Schiller 1995).
Cancer immunotherapy
Active Passive
Antigen Antigen Anti-tumor Adoptive cell

dependent independent monoclonal Ab transfer
Therapeutic Modulating T Enhancing immune

vaccines cell function cell function
GSK1572932A Immune checkpoint Cytokines: TCR engineering

Trastuzumab
TG4010 inhibitors: IL-2 CAR T
Cetuximab
Belagenpumatucel-L anti-CTLA-4 Ab IFN-a dendritic cells
Panitumumab
Tergenpumatucel-L anti-PD-1 Ab TILs infusion
Racotumomab anti-PD-L1 Ab
L-BLP25
Figure 1. Active and passive types of cancer immunotherapy in lung cancer. Ab = antibody; CTLA = cytotoxic T lymphocyte-
associated protein; PD-1 = programmed cell death protein-1; PD-L1 = programmed cell death ligand-1; IL-2 = interleukin-2;
IFN-α = interferon alpha; CAR-T = chimeric antigen receptor T cells; TIL = tumor-infiltrating lymphocytes.
Cancer Vaccines
Therapeutic cancer vaccines are designed to eliminate cancer cells by augmenting the
patient’s own immune responses. This type of vaccine contrasts with prophylactic vaccines,
which are usually administered to healthy individuals. Cancer vaccines can be categorized
into several major types, such as cellular vaccines, peptide vaccines, and genetic vaccines
(Cuppens 2014, Decoster 2012). Cellular vaccines can be either autologous or allogeneic.
Autologous tumor cell vaccines are developed by isolating tumor cells from an individual
patient, creating a vaccine formulation, and then administering the formulation back to
the same individual, usually in combination with an immune system-stimulating adjuvant
therapy. These vaccines were one of the first types of cancer vaccines tested and have
the advantage of potentially eliciting an immune response to a large spectrum of tumor-
associated antigens expressed by the patient’s own tumor, resulting in tumor destruction.
Although similar to autologous tumor cell vaccines, allogeneic vaccines are derived by taking
tumor cells from one patient, creating a vaccine formulation, and then administering the
formulation back to another patient with the same type of cancer.
CANCER IMMUNOTHERAPY FOR LUNG CANCER 23
Unlike cellular vaccines, which are made directly from patients’ tumors, peptide vac-
cines are often synthesized in vitro to mimic tumor-associated proteins, with the goal of
eliciting an immune response against tumor cells that express that specific tumor-associated
protein. Genetic vaccines are composed of synthetic DNA or RNA molecules that encode
for tumor-associated proteins and are administered either alone or packaged within a non-
pathogenic virus. The genetic material is taken up by cells within the recipient, translated
in the encoded proteins, processed, and presented to the immune system to provoke an
immune response against tumor-associated proteins.
Early studies of vaccine therapy with bacillus Calmette-Guerin in the adjuvant and neo-
adjuvant setting were negative (Bakker 1986, Miller 1982, Matthay 1986). In the modern era,
multiple vaccine studies have been conducted in early, locally advanced, and advanced-stage
NSCLC. The melanoma-associated antigen (MAGE)-A3 recombinant protein vaccine has
been extensively studied in the adjuvant setting after complete resection. A randomized
phase II study showed that, for patients with completely resected stage IB-II, MAGE-A3–
positive NSCLC who received no adjuvant chemotherapy, there was a trend toward superior
disease-free survival with MAGE-A3 vaccine compared with placebo after a median follow
up of 70 months (HR: 0.75; 95% CI: 0.46-1.23; p=0.254) (Vansteenkiste 2016). However, no
clinical benefit was found in the subsequent randomized, double-blind, placebo-controlled
large phase III study (MAGRIT) of completely resected stage IB-IIIA MAGE-A3 positive
NSCLC, with or without adjuvant chemotherapy. For the overall population in this later study,
the median disease-free survival was 60.5 months for the MAGE-A3 vaccine group and 57.9
months for the placebo group (HR: 1.02, 95% CI: 0.89-1.18; p=0.74). In the subgroup that did
not receive adjuvant chemotherapy, the median disease-free survival was 58.0 months for
the vaccine group and 56.9 months for placebo group (HR: 0.97; 95% CI: 0.80-1.18; p=0.76)
(Vansteenkiste 2016). Based on these results, the clinical development of the MAGE-A3
vaccine has been terminated.
Tecemotide (L-BLP25) is a peptide vaccine based on a 25-amino acid sequence from
the mucinous glycoprotein-1 (MUC1) protein that demonstrated promising activity in the
setting of locally advanced NSCLC in a phase II study (Butts 2005), subsequently result-
ing in the initiation of two randomized studies. One was a global phase III trial, START, in
which tecemotide was compared with placebo for patients with stage III NSCLC without
disease progression after chemoradiation therapy (Butts 2014). The second trial, INSPIRE,
was a randomized phase II trial of Asian patients (Wu 2011). The START trial showed no
difference in median overall survival between the tecemotide arm and placebo arms (25.6
months vs. 22.3 months; adjusted HR: 0.88; 95% CI: 0.75-1.03; p=0.123). However, follow-
ing a prespecified subgroup analysis, the median overall survival did differ between the
vaccine and placebo arms for patients who received concurrent chemoradiation therapy
(30.8 months vs. 20.6 months; HR: 0.78; 95% CI: 0.64-0.95; p=0.016) compared with patients
who received sequential chemoradiation therapy (19.4 months vs. 24.6 months; HR: 1.12;
95% CI: 0.87-1.44; p=0.38). INSPIRE was terminated in 2014 after Merck announced that it
planned to discontinue the clinical development of tecemotide as monotherapy for patients
with stage III NSCLC because of disappointing results from the Japanese phase I/II EMR
63325-009 study (Merck KGaA 2014).
In the advanced-disease setting, TG4010, another MUC1-targeting vaccine that uses

a viral vector to express both the full-length MUC1 and IL-2 (a T-cell stimulant), showed
promising clinical activity in the TG4010 immunotherapy and first-line chemotherapy
for advanced NSCLC (TIME) study. Results from the phase IIb part of the randomized,
double-blind, placebo-controlled, phase IIb/III trial showed that, in the overall population,
progression-free survival was 5.9 months for the TG4010 group and 5.1 months for the
placebo group (HR: 0.74; 95% CI: 0.55–0.98; p=0.019) (Quoix 2016). The phase III portion
of the trial is continuing.
Belagenpumatucel-L is an allogeneic whole tumor-cell vaccine derived from four radiated
NSCLC cell lines of varying histologies that also express an antisense transgene for trans-
forming growth factor beta2, which downregulates the immunosuppressant transforming
growth factor beta2. The findings of a phase II study suggested clinical efficacy in advanced
NSCLC (Nemunaitis 2006), and a phase III study (STOP) was initiated to randomly assign
patients with stage III/IV NSCLC in whom disease did not progress after platinum-based
chemotherapy to either belagenpumatucel-L or placebo (Giaccone 2015). There was no
significant difference in overall survival between the two treatment arms (20.3 months
vs. 17.8 months; HR: 0.94; p=0.594); likewise, there was no difference in progression-free
survival between the two groups (4.3 months vs. 4.0 months; HR: 0.99; p=0.947).
Epidermal growth factor receptor (EGFR) is an important signalling pathway in NSCLC,
and a vaccine has been developed against its cognate ligand EGF using recombinant human
EGF coupled to a carrier protein. In a randomized phase II study, patients with stage IIIB/IV
NSCLC were randomly assigned to receive either best supportive care or EGF vaccinations
after first-line chemotherapy (Neninger 2008). In the overall population, there was a trend
for improved overall survival, and a significant survival advantage for patients who had a
good antibody response to EGF. A later phase III trial included patients with stage IIIB/
IV NSCLC who were randomly assigned after first-line chemotherapy to either vaccine or
best supportive care. In the safety population, overall survival was 10.83 months for the
vaccine arm and 8.86 months for the control arm (Rodriguez 2016). This difference was not
significant according to the standard log rank (HR: 0.82; p=0.100), but was significant accord-
ing to a weighted log rank (p=0.04) that was applied once the nonproportionality of the
hazard ratio was verified. In the per-protocol setting (patients who received at least four
vaccine doses), overall survival differed significantly between the vaccine and best sup-
portive care arms (12.43 months vs. 9.43 months; HR: 0.77; p=0.036). In addition, overall
survival was longer (14.66 months) for vaccinated patients with high EGF concentrations
at baseline.
Immune Checkpoint Inhibitors

More recently, a deeper understanding of the interaction between the immune system and
tumors has led to the identification of CTLA-4 and PD-1/ PD-L1 as key factors by which
tumors evade host immune response (Pardoll 2012). This discovery has led to the devel-
opment of a new generation of immunotherapy agents that target these molecules. The
immune checkpoint inhibitors represent an important breakthrough in the treatment of
cancer. Multiple studies have shown immune checkpoint inhibitors to be highly active and
durable in a variety of solid tumors, including NSCLC. The immune checkpoint inhibitors
Table 1. Selected Inhibitory and Stimulatory Agents Targeting the Immune Checkpoint Pathway
Stage of Development
Target Agent in NSCLC Manufacturer
Inhibitory Agents
CTLA-4 Ipilimumab Phase III Bristol-Myers Squibb
Tremelimumab Phase III AstraZeneca/MedImmune
PD-1 Nivolumab (BMS936558) Approved by US FDA Bristol-Myers Squibb/ONO
Pembrolizumab (MK-3475) Approved by US FDA Merck
Pidilizumab (CT-011) Phase I-II Cure Tech/Teva
PDR001 Phase I-II Novartis
PD-L1 Atezolizumab (MPDL3280A) Approved by US FDA Genentech
Durvalumab (MEDI4736) Phase III AstraZeneca/MedImmune
Avelumab (MSB0010718C) Phase III Pfizer/Merck Serono
LAG3 LAG525 Phase I-II Novartis
KIR Lirilumab Phase I-II Bristol-Myers Squibb
Stimulatory Agents
OX40 MEDI0562 Phase I AstraZeneca/MedImmune
MEDI6383 Phase I AstraZeneca/MedImmune
MOXR0916 Phase I Genentech
4-1BB Utomilumab (PF-05082566) Phase I Pfizer
Urelumab (BMS- 663513) Phase I-II Bristol-Myers Squibb
GITR MK-4166 Phase I Merck
CTLA-4 = cytotoxic T lymphocyte-associated protein 4; PD-1 = programmed cell death protein-1; PD-L1 = programmed
cell death ligand-1; FDA = Food and Drug Administration; LAG3 = lymphocyte-activation gene 3; KIR = killer-cell
immunoglobulin-like receptor; GITR = glucocorticoid-induced tumor necrosis factor receptor.
Bristol-Myers Squibb, New York, USA; AstraZeneca, Cambridge, UK; MedImmune, Gaithersburg, Maryland, USA; ONO
Pharmaceutical Co., LTD, Osaka, Japan; Merck, Kenilworth, New Jersey, USA; Cure Tech, Yavne, Israel’ Teva Pharmaceutical
Industries, Ltd, Petach Tikva, Israel; Novartis International AG, Basel, Switzerland; Genentech, South San Francisco, USA,
Pfizer Oncology, New York, USA; Merck Serono, Darmstadt, Germany.
developed include the CTLA-4 inhibitors, PD-1 inhibitors, and PD-L1 inhibitors (Table 1).
Other immune checkpoint inhibitors under development include lymphocyte-activation
gene 3 (LAG3) and killer-cell immunoglobulin-like receptor inhibitors, as well as immune
checkpoint stimulatory agents such as agonists to OX40, 4-1BB, and GITR (Sundar 2014).
Further discussion on these latter agents, however, is beyond the scope of this chapter and
they have been reviewed elsewhere (Pardoll 2012).
CTLA-4 Inhibitors
Ipilimumab is a recombinant human IgG1 monoclonal antibody that binds to CTLA-4 and
prevents the downregulation of T-cells at early stages of T-cell activation (Figure 2). Activity
of ipilimumab in advanced melanoma has been clearly demonstrated in two large phase
III trials (Hodi 2010, Robert 2011), which resulted in US Food and Drug Administration
(FDA) approval in 2011. Ipilimumab produces durable long-term survival, as demonstrated
by a significantly longer 5-year survival rate for ipilimumab plus dacarbazine compared
with placebo plus dacarbazine (18.2% [95% Activation

CI: 13.6% to 23.4%] vs. 8.8% [95% CI, 5.7% to CD80/
IL-2
CD83 CD28
12.8%]; p=0.0002) (Maio 2015). APC T-Cell
IFN-γ
Ipilimumab in combination with chemo-
+
therapy has been studied in patients with
advanced-stage NSCLC who had not received CD40 CD40L
prior treatment. In this three-arm phase II

study, patients were randomly assigned to che-
motherapy (carboplatin plus paclitaxel) alone, CTLA-4 Checkpoint
chemotherapy with phased ipilimumab, or che- CD80/ IL-2
CD83 CTLA-4
motherapy with concurrent ipilimumab. The
CD28
primary endpoint of the study was immune- IFN-γ
related progression-free survival, which was
4.6 months for the chemotherapy alone arm,
5.7 months for the phased ipilimumab arm (HR: CD40
0.72; p=0.05) and 5.5 months for the concur-

Anti-CTLA-4
rent ipilimumab arm (HR: 0.81; p=0.13) (Lynch
2012). Subset analysis demonstrated that the T-Cell
Reinvigoration Anti-CTLA-4
immune-related progression-free survival
CTLA-4
in the phased ipililumab arm was longer for CD80/
IL-2
CD83 CD28
patients with NSCLC of squamous histology
IFN-γ
than for patients with NSCLC of nonsquamous
+
histology. To confirm these results, a larger
phase III trial (NCT02279732) has been initi- CD40 CD40L
ated for patients with squamous cell NSCLC.

IL-12
PD-1 Inhibitors Figure 2. The anti-cytotoxic T lymphocyte-associated pro-

tein-4 (CTLA-4) antibodies, such as ipilimumab and tremeli-
The PD-1 inhibitors include agents such as lumab, blocking the inhibitory interaction between CTLA-4
nivolumab and pembrolizumab. Nivolumab is a and B7, thus resulting in T-cell reinvigoration.
fully human immunoglobulin G4 (IgG4) mono-
clonal antibody that disrupts PD-1–mediated signalling, thus releasing T-cells from their
inhibitory interaction with PD-L1 and PD-L2 (Figure 3). Pembrolizumab is a humanized high-
affinity, IgG4/kappa isotype monoclonal antibody that also blocks PD-L1 ligating with PD-1
on T-cells, resulting in the activation of tumor-specific cytotoxic T-cells. Pembrolizumab
has an optimized fragment crystallizable region to minimize antibody-dependent cellular
cytotoxicity (ADCC) and complement-dependent cytotoxicity (CDC) (Homet 2015). This
action may be important because an intact ADCC has the potential to cause a depletion of
activated T-cells and tumor-infiltrating lymphocytes and result in diminished activity, as
PD-1 is expressed on T effector cells and other immune cells (Chen 2012).
Nivolumab
Nivolumab was first noted in early-phase studies to be active in melanoma (Topalian 2012).
The findings of subsequent phase III studies confirmed the superiority of nivolumab over
standard of care, leading to its approval by the FDA for the treatment of melanoma. More
Activated CTL recently, the combination of nivolumab and ipi-

Perforin Granzyme limumab was approved as first-line treatment
IFN-γ
IL-2
of patients with advanced-stage melanoma,
T-Cell regardless of BRAF V600E status (Larkin 2015).
CTL
Nivolumab is also active in a variety of solid
FASL FAS tumors and has been approved by the FDA for
TRAIL TRAILR the management of advanced NSCLC, advanced
PD-1 PD-L1 renal cell carcinoma, and classical Hodgkin
lymphoma (Motzer 2015, Ansell 2015).
PD-L1
Checkpoint IL-2 Activity in advanced-stage NSCLC
Exhausted In a phase Ib study of patients with selected
CTL
advanced-stage solid tumors, including NSCLC,
patients were treated with escalating doses of
PD-1 PD-L1 nivolumab. Within the NSCLC cohort (129
Anti-PD-1 patients), the objective response rate was
or
Anti-PD-L1 17%, with a median duration of response of
PD-1/PD-L1 74 weeks (range, 6.1–133.9 weeks). Of note,
Checkpoint- many patients were heavily pretreated, with
blocking mAb
Perforin
Granzyme 54% having received at least three prior lines
IL-2 of therapy (Topalian 2012). At longer follow-
Reinvigorated
CTL
up, the median overall survival across all doses
of nivolumab was 9.9 months; and the 1-, 2-,
FASL FAS and 3-year survival rates were 42%, 24%, and
TRAIL TRAILR 18%, respectively (Gettinger 2015) (Table 2).
PD-1 PD-L1 Nivolumab was well tolerated. Forty-one per-
Anti-PD-L1
cent of patients with NSCLC experienced any
Anti-PD-1
treatment-related adverse events; grade 3 or 4
Figure 3. The anti-programmed cell death protein-1
toxicities developed in 4.7%. Based on the find-
(PD-1) antibodies, such as pembrolizumab and nivolumab,
inhibit the interaction between PD-1 and its ligands pro- ings of this study, the recommended dose for
grammed cell death ligand-1 and 2 (PD-L1 and PD-L2). nivolumab was 3 mg/kg every 2 weeks.
The anti–PD-L1 inhibitors (atezolizumab, durvalumab,
Single-arm trials have demonstrated
and avelumab) block PD-L1 from ligating with PD-1. The
aim of anti–PD-1 and anti–PD-L1 treatment is to block the effectiveness of nivolumab in manag-
the inhibitory signalling resulting from PD-1 ligation with ing advanced-stage NSCLC (Table 2). In
PD-L1, thus restoring cytotoxic T-cell activity.
a phase II single-arm trial (CheckMate
063) of nivolumab for patients with squa-
mous cell NSCLC who were treated with third-line therapy and beyond, the partial
response rate was 14.5%, and 26% of patients had stable disease (Rizvi 2015A). The
overall survival was 8.2 months, and the 1-year survival was approximately 41%
(Table 2). Of note, the study population was highly refractory to treatment, with 65% of
patients treated with at least three prior lines of systemic therapy. In addition, 61% of patients
had disease progression as the best response to the most recent therapy. In a phase II Japanese
study (ONO-4538-05), the objective response rate was 25.7% for patients with squamous
cell NSCLC and 19.7% for patients with nonsquamous cell NSCLC (Takeda 2015). In another
Table 2. Clinical Outcomes for Patients with Advanced NSCLC Treated with Anti–PD-1/PD-L1 Immune
Checkpoint Inhibitors
Trial Treatment Histologic ORR Progression-free

Study Phase Line Subtype Drug (%) Survival Overall Survival
At 24 At 1 At 1
No. of Mos. Weeks Year No. of Mos. Year
(95% CI) (%) (%) (95% CI) (%)
PD-1 inhibitors
Gettinger Phase Second or Sq and Nivolumab 17.1 2.3 (1.8-3.7) 33 22 9.9 (7.8-12.4) 42
2016 I more nonaq
Rizvi Phase Third or Sq Nivolumab 14.5 1.9 (1.8–3.2) 25.9 20 8.2 (6.1-10.9) 41
2015a II more (partial)
Takeda Phase Second Sq and- Nivolumab 25.7 4.2 (1.5-7.1) NR NR Not reached NR
2015 II nonsq (12.4-not
reached)
Hussein Phase Second or Sq and Nivolumab 12 NR NR NR NR NR
2015 II more nonsq
Brahmer Phase Second Sq Nivolumab 20 3.5 (2.1-4.9) NR 21 9.2 (7.3-13.3) 42
2015 III
Docetaxel 9 2.8 (2.1 -3.5) NR 6 6.0 (5.1-7.3) 24
Borghaei Phase Second Nonsq Nivolumab 19 2.3 NR 19 12.2 (9.7-15.0) 51

2015 III
Docetaxel 12 4.2 NR 8 9.4 (8.1-10.7) 39
Socinski Phase First Sq and Nivolumab 26.1 4.2 NR 23.6 14.4 56.3
2016 III nonsq
Platinum 33.5 5.9 NR 23.2 13.2 53.6
doublet
Garon Phase Any Sq and Pembrolizumab 19.4 3.7 (2.9-4.1) NR NR 12.0 (9.3-14.7) NR
2015 I nonsq
Herbst Phase Second or Sq and Pembrolizumab 18 3.9 NR NR 10.4 (9.4-11.9) 43.2
2016 II/III more nonsq (2 mg/kg)
Pembrolizumab 18.5 4.0 NR NR 12.7 (10.0-17.3) 52.3
(10 mg/kg)
Docetaxel 9.3 4.0 NR NR 8.5 (7.5-9.8) 34.6
Reck Phase First Sq and Pembrolizumab 44.8 10.3 62.1 NR Not reached NR
2016 III nonq
Platinum 27.8 6.0 50.3 NR Not reached NR
doublet
PD-L1 Inhibitors
Herbst Phase Any Sq and Atezolizumab 23 15 44.7 NR NR NR
2014 I nonsq weeks
Spigel Phase Second or Sq and Atezolizumab 16 2.7 32 NR 10.6 (5.8-NR) 48
2015 II more nonsq
Fehrenbacher Phase Second or Sq and Atezolizumab 14.6 2.7 NR NR 12.6 NR
2016 II third nonsq
Docetaxel 14.7 3.0 NR NR 9.7 NR
continued on next page

Barlesi Phase Second or Sq and Atezolizumab 14 2.8 NR NR 13.8 55

2016 III third nonsq
Docetaxel 13 4.0 NR NR 9.6 41
Rizvi Phase Any Sq and Durvalumab 14 NR NR NR PD-L1+: not NR
2015 I nonsq reached
PD-L1-ve: 8.9
Verschraegen Phase First Sq and Avelumab 18.7 11.6 35.6 NR NR NR
2016 Ib nonsq weeks
ORR = objective response rate; Sq = squamous; Nonsq = nonsquamous; NR = not reported.
phase II study (CheckMate 153), 824 patients with advanced NSCLC were treated for 1 year
with nivolumab. The partial response and stable disease rates were 12% and 44%, respec-
tively. Responses were independent of PD-L1 expression (Hussein2015).
Second-line nivolumab was superior to docetaxel in two subsequent randomized phase
III trials of patients with advanced NSCLC who received a platinum-based chemotherapy
doublet (Table 2). In a study of 272 patients with squamous cell NSCLC (CheckMate 017),
the median overall survival and 1-year survival was better for nivolumab than for docetaxel
(Table 2). The hazard ratio for death was 0.59 with nivolumab (p<0.001) (Brahmer 2015).
In the other study (CheckMate 057), which included patients with advanced nonsquamous
cell NSCLC, second-line nivolumab was also associated with better overall survival and
1-year survival than docetaxel (HR: 0.73) (Borghaei 2015) (Table 2). In subset biomarker
analysis, with increasing PD-L1 expression ≥1%, ≥5% and ≥10%, there was improvement
in PFS with a HR of 0.70, 0.54 and 0.52, respectively and in OS with a HR of 0.58, 0.43 and
0.40, respectively. Conversely, in tumors with low PD-L1 expression of < 1%, < 5%, and
< 10%, the HR for PFS was 1.19, 1.31, and 1.24, respectively and for OS was 0.87, 0.96, and
0.96, respectively.
CheckMate 017 is the first study to show the beneficial effect of immune checkpoint inhib-
itors as assessed by patient-reported outcomes. At week 12, 20.0% of patients who received
nivolumab and 21.9% of patients who received docetaxel had symptom improvement as
assessed by the Lung Cancer Symptom Scale (Gralla 2015). Patients who received nivolumab
had greater symptom improvement compared with patients treated with docetaxel. The
time to first disease-related deterioration as assessed by Lung Cancer Symptom Scale Global
Health Related Quality of Life was longer in the nivolumab arm than in the docetaxel
arm (HR: 0.58; 95% CI: 0.39-0.86). Patient-reported outcomes, as measured by the scores
on EQ-5D and EQ-5D VAS, improved in the nivolumab arm, with the time to first dis-
ease-related deterioration on the EQ-5D index favoring patients treated with nivolumab
(Reck 2015).
The safety and efficacy of single-agent nivolumab in the first-line treatment of patients
with advanced NSCLC was reported in CHECKMATE 012. Adverse events occurred in 71%
of patients, with the most common adverse events being fatigue (29%), rash (19%), nausea
(14%), diarrhea (12%), pruritus (12%), and arthralgia (10%). The confirmed overall response
rate was 23%, and the progression-free survival and overall survival were 3.6 months and
19.4 months, respectively. The 24-week progression-free survival rate was 41%, and the
1-year overall survival rate was 73% (Gettinger 2016). More recently, in a phase III study
of first-line nivolumab compared with a platinum-based chemotherapy doublet for tumors
with PD-L1 expression of 5% or greater (CheckMate 026), the progression-free survival was
longer for the chemotherapy arm but overall survival was better for the nivolumab arm
(Socinski 2016). The objective response rate was lower for the nivolumab arm (Table 2).
Activity in SCLC
SCLC is most often extensive-stage disease at the time of diagnosis. Although first-line
platinum-based chemotherapy doublets have activity, disease inevitably progresses, and
response rates in the second-line setting are low and not durable. The activity and safety
of nivolumab with or without ipilimumab in previously treated SCLC were evaluated in
CheckMate 032. The objective response rate was 10% with 3 mg/kg of nivolumab alone,
23% with 1 mg/kg of nivolumab in combination with 3 mg/kg of ipilimumab, and 19% with
3 mg/kg of nivolumab in combination with 1 mg/kg of ipilimumab (Antonia 2016A). PD-L1
expression was not associated with responses.
Pembrolizumab
Pembrolizumab is active in a variety of solid tumors including melanoma, mismatch repair-
deficient colorectal cancer, NSCLC, gastric cancer, and urothelial cancer, as well as in Merkel
cell and Hodgkin lymphoma (Robert 2015, Le 2015, Muro 2016, Seiwert 2016, Nghiem 2016,
Armand 2016). The agent has been approved by the FDA for the treatment of metastatic
melanoma, advanced-stage NSCLC, and recurrent or metastatic head and neck squamous
cell carcinoma.
Activity in NSCLC
The efficacy and safety of pembrolizumab at two different doses in patients with untreated
or previously treated advanced-stage NSCLC was reported in KEYNOTE-001, a large phase
I study. Among all patients, the objective response rate was 19.4%, and the median duration
of response was 12.5 months. The progression-free survival was 3.7 months, and overall
survival was 12.0 months (Garon 2015). The objective response rate was 18% among pre-
viously treated patients and 24.8% among untreated patients. For patients with a tumor
proportion score of at least 50%, the objective response rate was 45.2%, and progression-free
survival was 6.3 months. The objective response rate was similar regardless of dose, sched-
ule, and histologic subtype; the response rate was higher among smokers than nonsmokers.
Treatment-related adverse events of any grade occurred in 70.9% of patients; 9.5% had an
adverse event of grade 3 or higher.
Pembrolizumab was evaluated in a phase II/III study of patients with previously treated
advanced NSCLC (KEYNOTE-010). A total of 1,034 patients were randomly assigned to
either 2 mg/kg or 10 mg/kg of pembrolizumab or to 75 mg/m 2 of docetaxel every 3 weeks
(Herbst 2016). All patients had at least 1% of tumor cells that stained positively for PD-L1
protein expression on immunohistochemistry (IHC). The overall survival was improved
with both doses of pembrolizumab compared with docetaxel (Table 2). Among patients
with at least 50% of tumor cells expressing PD-L1, the overall survival was 14.9 and 17.3
months with pembrolizumab at doses of 2 mg/kg and 10 mg/kg, respectively, compared
with 8.2 months with docetaxel. Any grade of treatment-related adverse events occurred
in 63% of patients who received 2 mg/kg of pembrolizumab and in 66% of patients who
received the 10 mg/kg dose. Treatment-related toxicity was higher (81%) in the docetaxel
arm. Grade 3 to 5 treatment-related adverse events were less common among patients treated
with pembrolizumab (2mg/ kg, 13%; 10 mg/kg, 16%) compared with docetaxel (35%).
The safety and efficacy of first-line pembrolizumab in patients with advanced NSCLC
was evaluated in KEYNOTE-001. The progression-free and overall survival were 6.2 months
and 22.1 months, respectively. Increased PD-L1 expression was associated with longer
survival; in patients with PD-L1 expression of 50% or greater, progression-free and overall
survival were 12.5 months and not reached, respectively. In contrast, for tumors with PD-L1
expression of 1% to 49%, the progression-free and overall survival were 4.2 months and
14.7 months, respectively (Hui 2016).
In the phase III study for first-line therapy of advanced NSCLC, KEYNOTE-024, patients
with tumor PD-L1 expression of 50% or greater were randomly assigned to pembrolizumab
or a platinum-based chemotherapy doublet, and progression-free survival was significantly
better for pembrolizumab (HR: 0.50; 95% CI: 0.37–0.68; p<0.001) (Reck 2016). The hazard
ratio for overall survival was 0.60 (95% CI: 0.41–0.89; p=0.005). In addition, the response
rate was higher for pembrolizumab than for chemotherapy (Table 2), and fewer adverse
events were associated with pembrolizumab. These results are groundbreaking because this
study is the first to demonstrate the superiority of anti–PD-1 therapy over platinum-based
combination chemotherapy in the first-line setting for advanced NSCLC. Patients had no
sensitizing EGFR mutations or ALK translocations and had high PD-L1 expression.
Activity in SCLC
Preliminary data from a phase Ib multicohort study of pembrolizumab in patients with
previously treated PD-L1-positive SCLC include an objective response rate of 25% and a
disease-control rate of 31% (Ott 2015).
PD-L1 Inhibitors
PD-L1 inhibitors also obstruct PD-1/PD-L1 interactions but leave the PD-1/PD-L2 pathway
intact (Figure 3). The PD-L1 inhibitors include atezolizumab, durvalumab, and avelumab
(Table 1). Atezolizumab and durvalumab are human IgG1 anti–PD-L1 antibodies engineered
with mutations in their Fc domains to remove both ADCC and CDC activity. Avelumab is
a fully human IgG1 anti–PD-L1 monoclonal antibody and, unlike the other PD-1/PD-L1
inhibitors, it has appeared to have retained its ADCC and CDC activity in preclinical stud-
ies (Boyerinas 2015).
Several PD-L1 inhibitors have reported shown promising activity in Merkel cell carci-
noma, urothelial cancer, and NSCLC (Rosenberg 2016, Kaufman 2016, Massard 2016). Phase
III studies confirming the activity of these agents in various solid tumors are ongoing.
Atezolizumab
Atezolizumab was reported to be active in urothelial cancer in a phase I study (Powles 2014)
and was subsequently approved by the FDA for the treatment of advanced-stage urothelial
carcinoma. In a single-arm phase II study (IMvigor 210 trial) the objective response rate
was 16%, irrespective of immune cell PD-L1 expression, and was 28% for patients with 5%
or greater PD-L1 expression (Rosenberg 2016).
Atezolizumab was reported to be active in a phase I study of advanced-stage NSCLC

(Herbst 2014). In a dose-escalation and expansion study, the objective response rate was
23%, progression-free survival was 4 months, and overall survival was 16 months for patients
who received 20 mg/kg intravenously every 3 weeks (Horn 2015). In a randomized phase
II study (POPLAR) of patients who had received platinum-based chemotherapy, atezoli-
zumab was associated with superior overall survival (HR: 0.73; 95% CI: 0•53–0•99; p=0.04)
(Fehrenbacher 2016) (Table 2). In another phase II study (BIRCH), patients with advanced
NSCLC who were selected for PD-L1 expression received atezolizumab as first-line or sub-
sequent therapy. The objective response rates ranged from 17% to 27% (Besse 2016), and the
median overall survival was 14 months for patients who received atezolizumab as first-line
therapy. Overall survival has not yet been reached for patients who received atezolizumab
as subsequent therapy (Broderick 2016). The overall response rates ranged from 16% to
26% in a phase II study of a population with advanced NSCLC enriched for PD-L1 expres-
sion in tumor and immune cells (Spigel 2015). In the OAK trial, a phase III study of patients
with advanced, previously treated NSCLC who were randomly assigned to atezolizumab or
docetaxel, the overall survival was significantly better for atezolizumab (13.8 months vs.
9.6 months; HR: 0.73; 95% CI: 0.62-0.87; p=0.0003) (Rittmeyer 2017). The OAK study led
to FDA approval of atezolizumab for second-line therapy of advanced NSCLC.
Durvalumab
In a phase I/II study of durvalumab in the first-line setting in patients with NSCLC regard-
less of PD-L1 status. The overall response rate was 27% and the response rate was 29% in
PD-L1 positive tumors (defined as ≥ 25% of tumor cells expressing PD-L1) and 11% in PD-L1
negative tumors. (Antonia 2016B). In another phase I study of patients with previously
treated advanced-stage NSCLC, the response rate was 14% overall and 23% for patients with
PD-L1 expression (Rizvi 2015B). In a phase II study of patients with advanced NSCLC who
had received at least two prior lines of systemic therapy, activity was highly encouraging
(Figure 4). The objective response rate and 1-year survival rate increased according to PD-L1
A B
Figure 4. Hepatic metastases (arrows) from patient with non-small cell lung cancer at (A) baseline and
(B) after 7.5 months of treatment with durvalumab.
expression: 7.5% (less than 25% PD-L1 expression), 16.4% [(more than 25% expression)],
and 30.9% (more than 90% expression); the corresponding 1-year survival rates were 34.5%,
47.7%, and 50.8% (Garassino 2016).
Avelumab
The findings of early studies of avelumab in NSCLC have been promising, with an over-
all response rate of 12% for patients who had disease progression after platinum-based
chemotherapy. There was a trend toward greater activity in patients with PD-L1-positive
tumors (Gulley 2015). Among patients treated with avelumab in the first-line setting, the
objective response rate and disease-control rates were 18.7% and 64.0%, respectively
(Verschraegen 2016).
Combined CTLA-4 and PD-1/PD-L1 Inhibitors

CTLA-4 inhibitors are also being studied in combination with PD-1 and PD-L1 inhibitors.
The results of preclinical studies indicate that the combination may work synergistically
to produce enhanced antitumor activity (Curran 2010). The combination of ipilimumab
with nivolumab in advanced melanoma resulted in improved antitumor activity compared
with single-agent therapy; however, toxicities were increased with the combination therapy
(Larkin 2015).
Nivolumab has been combined with ipilimumab for advanced NSCLC in the first-line
setting in a phase I study (CheckMate 12). Results have included objective response rates
ranging from 13% to 39% (Hellman 2016). A randomized phase III trial (CheckMate 227) is
currently ongoing to compare nivolumab plus ipilimumab with nivolumab alone, nivolumab
plus platinum-based chemotherapy, and platinum-based chemotherapy alone in PD-L1-
defined untreated NSCLC.
Additionally, durvalumab has been combined with the CTLA-4 inhibitor tremelimumab
in a phase Ib trial of patients with advanced NSCLC. Although numerous adverse events
occurred during the dose-escalation phase of the study, antitumor activity was evident
(objective response rate of 23%), regardless of PD-L1 status in the evaluable patients in the
dose-expansion phase of the study (Antonia 2016A).
Detection of PD-L1 Expression

Benefit is found in only a subset of patients with NSCLC treated with PD-1/PD-L1 inhibi-
tors. As the PD-1/PD-L1 pathway is involved in immune escape in NSCLC, tumor PD-L1
expression with IHC has been used to identify patients who may benefit from PD-1/PD-L1
inhibitors. Studies of pembrolizumab have used PD-L1 IHC 22C3 PharmDx (Dako) for detec-
tion of tumor PD-L1 expression (Garon 2015, Herbst 2016). This test has been approved
as a companion diagnostic test by the FDA. Studies of nivolumab have used the anti-PD-
L1 antibody [28-8] (Abcam) (Borghaei 2016, Brahmer 2015), which has been approved as
a complementary diagnostic test. (The differences between companion and complementary
diagnostic tests are discussed in Chapter 10). The antibody clones SP142 and SP263 have
been used to develop tumor PD-L1 expression in studies of atezolizumab (Fehrenbacher
2016) and durvalumab (Rebelatto 2016), respectively. The use of PD-L1 as a biomarker
is complex, given that immune-cell PD-L1 expression Table 3. Antibody Clones in Immuno-
has been associated with clinical outcome in studies of histochemistry Diagnostic Assays for
Programmed Death Ligand 1 (PD-L1)
atezolizumab (Herbst 2014, Fehrenbacher 2016). The
plethora of companion diagnostics developed for each Drug Antibody Clone
PD-1/ PD-L1 inhibitor has created challenges, as these Nivolumab 28-8 (Dako)
assays include different IHC antibody clones, staining
protocols and platforms, scoring systems, and cutoffs Pembrolizumab 22C3 (Dako)
for defining positivity (Table 3) (Sholl 2016, Kerr 2015). Atezolizumab SP142 (Ventana)
Each IHC antibody clone and the association between
Durvalumab SP263 (Ventana)
PD-L1 expression and clinical outcome are discussed
in more detail in Chapters 4-8. Avelumab 73-10 (Dako)
Conclusion
The PD-1 and PD-L1 immune checkpoint inhibitors herald a new therapeutic paradigm in
NSCLC, with benefit demonstrated in multiple studies for patients with previously treated
advanced disease. According to recent data, first-line pembrolizumab has greater effi-
cacy than a platinum-based chemotherapy doublet in a selected patient population and
has received FDA approval for first-line treatment. Studies to examine combining PD-1 or
PD-L1 inhibitors with chemotherapy, targeted therapy, or with other immunotherapeutic
agents are ongoing. PD-L1 expression is associated with improved efficacy of checkpoint
inhibitors, but this expression is an imperfect biomarker. Furthermore, the advent of com-
panion diagnostics developed for each PD-1/ PD-L1 inhibitor has created challenges for
pathologists and oncologists.
Immunohistochemistry for PD-L1
By Erik Thunnissen, Yasushi Yatabe, Sylvie Lantuéjoul, and Lukas Bubendorf 3
Immunohistochemistry (IHC) is a technique that allows visualization of proteins in
histologic sections, and a similar approach on cells in cytologic specimens is called immu-
nocytochemistry. With IHC, the variable domain of the primary antibody recognizes and
binds to the three-dimensional structure of a protein, an epitope, present in the section. A
second antibody that binds to the primary antibody and subsequent chemical reactions are
used to visualize the localization of the epitope, a process known as signal enhancement.
The location of the IHC staining is detected in the tissue context with use of a microscope.
IHC staining may be located on or in one or more subcellular areas, such as on the cell
membrane, in the cytoplasm, or in the nucleus. IHC is a rapid and relatively inexpensive
method that is preferred by most pathologists primarily because it allows for the evaluation
of tissue architecture and tumor cells.
Potentially, IHC can detect rare positive cells more easily than fluorescence in situ hybrid-
ization, even at a low magnification, because of the high contrast of IHC-positive tumor cells
in an IHC-negative background. IHC can be performed successfully on a variety of tumor
specimens—formalin -fixed paraffin-embedded (FFPE) tissue blocks, fluid, and fine-needle
aspiration cytology cell blocks or smears--as long as at least a few clusters of viable tumor
cells are present in the specimen. In addition, a validated and robust programmed cell death
ligand-1 (PD-L1) IHC assay provides a cost-effective platform. Several methodologic aspects
may influence the outcome of IHC, which are discussed here, in general and specifically
for PD-L1.
Preanalytic Phase
General
FFPE tissue is the global standard material for IHC. The cold ischemia time—the time
between sampling of the tissue and the start of fixation—should be kept to a minimum to
avoid cold ischemia effects. Regardless of origin, diagnostic biopsy or surgical resection
specimens should immediately be fixed in an adequate amount (ratio of at least 10 times
the volume of the tissue specimen) of 10% neutral buffered formalin (ie, 4% formaldehyde),
processed, and embedded in paraffin (FFPE tissue). Fixation times of fewer than 6 hours
are not recommended because conventional H&E staining, as well as IHC, can be adversely
affected. For practical purposes, a fixation time of 6 to 48 hours is recommended for all
specimens (Thunnissen 2012, Kerr 2014).
During gross handling of resection tissue, samples should be cut to slices 3 or 4 mm thick.
The cassettes containing the tissue samples should also be placed in buffered formalin, then
dehydrated and cleared in a series of alcohols and xylene, followed by infiltration with melted
paraffin. The paraffin temperature should not exceed 60°C. The FFPE samples are assumed
to be stable and preserved against oxidation. FFPE tissue specimens are cut to create sec-
tions of 3 to 4 μm, which are mounted and dried onto glass slides, ready for staining. After
sectioning, the tissues are mounted on special coated slides suitable for IHC. If not used
within a few days, the tissue sections should be stored in the dark, in a closed box at 2°C to
8°C to preserve antigenicity, and stained within 3 months of sectioning to avoid possible
false-negative results. These measures aim to prevent photo-oxidation and drying, which
lead to loss of antigenicity (Blind 2008). An alternative procedure involves dipping glass
slides with mounted sections in paraffin wax to avoid oxidative damage. When such slides
are subsequently used for IHC, care must be taken to use enough solvent to dissolve all the
excess paraffin wax. It is possible for inadequate volumes of solvent to become saturated,
leading to incomplete dewaxing. Incomplete wax removal hampers epitope retrieval and
IHC staining. In addition, sections may detach from the glass slides during dewaxing.
Decalcification can destroy antigenicity, especially when highly acidic agents, such as
hydrochloric acid and nitric acid, are used. Weaker acids, such as EDTA, an effective decal-
cifying agent, have become more popular because they do not seem to interfere with IHC
for breast cancer testing (Schrijver 2016).
For other organs and proteins, some generalizable effects are known. Regarding breast
and colorectal cancers, a rapid change in phosphoepitope-specific protein staining was
found to be related to delay in fixation (Bonnas 2012, Meric-Bernstam 2014, Theiss 2014,
Vassilakopoulou 2015). Frequently, staining signals—for example, for expression of phos-
phorylated Akt (pAkt)—were substantially reduced or lost in resection specimens compared
with small biopsy specimens (Vassilakopoulou 2015). Thus, for phosphoproteins, which
are usually less stable than other proteins, a delay in fixation of 30 minutes to 1 hour can
negatively affect measurement outcome. In the cases of HER2, estrogen and/or progesterone
receptors, and Ki67 IHC in breast cancer, cold ischemia of up to 3 to 4 hours was shown to
have no deleterious effect (Neumeister 2012, Portier 2013, Li 2013).
PD-L1
There are several recommendations regarding preanalytic conditions for preparation and
storage of tissues for future PD-L1 staining (Table 1). For PD-L1 there is, to date, no infor-
mation about possible effect of fixation delay. Proper tissue collection and processing may
reduce the failure rate. Currently, an ischemia time from excision to formalin fixation start
time of fewer than 30 minutes followed by immersion in neutral buffered formalin for 6
to 48 hours is recommended. Biopsy specimens should be immersed immediately in buff-
ered formalin. According to the interpretation manuals for the 22C3 and 28-8 pharmDx
kits (Agilent Technologies/Dako), the only critical step for PD-L1 IHC is a fixation of at
IMMUNOHISTOCHEMISTRY FOR PD-L1 37
Table 1. Recommended Preanalytic Conditions for least 3 hours. The other parameters
Immunohistochemistry (IHC)
(ie, surfixation, paraffin embed-
Parameter Recommendation
ding, dehydration, and sectioning)
Cold ischemia time Fewer than 30 minutes if possible,
should follow the standard pro-
not exceeding 1 hour
cedure. Sample age and storage
Fixative 10% neutral buffered formalin
after sectioning also are known to
Time of fixation (biopsy) 6 to 48 hours
affect staining results; the details
Time of fixation (resection) 24 to 48 hours are described further in this chap
Preparation Paraffin-embedded sections, cut ter. For unstained slides, staining
at a thickness of 3 to 5 μm
for PD-L1 within 2 months is rec-
Specimen storage Tissue blocks ommended. It is acknowledged,
Storage time for blocks Fewer than 3 years for PD-L1 IHC however, that the diagnosis is
Storage conditions for Prevented from light, heat, and usually not known at the time of
blocks humidity sample excision, and the sample
Storage time for cut Variable, depending on storage will be required for all necessary
sections conditions. Refer to assay package diagnostic steps not just PD-L1
insert.
IHC. Biopsy tissue for the sole
Decalcification EDTA, if necessary
purpose of PD-L1 IHC is likely to
PD-L1 = programmed cell death-ligand 1.
be rare; samples cannot usually be
specifically fixed and processed to suit one particular biomarker test. For unstained slides,
it is best to stain them immediately after they are cut. Epitope stability may be affected by
prolonged storage but this will depend on the time and conditions of storage. Manufacturers
have different recommendations and readers are referred to the respective package inserts.
Manufacturers of PD-L1 assays emphasize that their assays have not been validated for
decalcified tissue. Thus, PD-L1 IHC on decalcified tissues should be avoided when other
tissue is available or should be interpreted with great caution until further validation studies
on PD-L1 IHC have become available.
Specimen age for PD-L1 testing should be fewer than 3 years, as in one study, the age
of the blocks was associated with the prevalence of PD-L1. The prevalence was similar
among samples that were less than months old, 3 months to 1 year old, and 1 to 3 years old
(approximately 30%) and substantially lower among samples that were more than 3 years
old (13%) (Midha 2016).
Overall, the tissue-handling procedure for PD-L1 should be the same as for other diag-
nostic or predictive markers, such as ALK (Cree 2016).
Analytic Phase
General
Several issues must be controlled and optimized for during the analytical procedure: the
development of adequate antibodies, epitope retrieval, type and concentration of the anti-
body, incubation time, incubation temperature, and signal enhancement (eg, with a tyramide
cascade and intercalation of an antibody-enhanced polymer).
Antigen preservation for IHC is epitope dependent, and some epitopes may not be ham-
pered by fixation times of as long as 120 hours. Once fixation is started, the target protein
for the IHC assay may inadvertently change its shape (3-dimensional configuration) due
to the fixation. In practice, neutral buffered formaldehyde is used for fixation, giving rise
to cross-links between proteins and stabilization of the tissue. Therefore, the target pro-
tein conformation may be different between frozen and fixed tissue. The cross-links could
possibly hamper the epitope recognition by the primary antibody. To this end, different
epitope-retrieval buffers are tested, and the optimal buffer is chosen for the standard stain-
ing procedure. A variety of different antigen unmasking and retrieval steps may be used in
different laboratories. Care should be taken to use only a recommended and proven method
for the PD-L1 IHC assay in use. It should be noted that none of the assays have been validated
for use on decalcified tissue specimens and therefore, great care should be taken if these
samples are used for PD-L1 testing.
PD-L1
For the actual PD-L1 testing, the same issues must be controlled for and optimized. For
commercial tests, these issues have been standardized and clinically validated (Table 2).
(These issues are also discussed in Chapters 4 to 8). The preanalytic conditions may affect
staining outcome, as previously discussed. For laboratory-developed tests (LDTs), adequate
clinical validation, as well as essential analytic validation, is of crucial importance. (LDTs
are discussed further in this chapter, as well as in Chapter 9.)
Table 2. Programmed Cell Death Ligand 1 (PD-L1) Immunohistochemistry Assays According to Drugs
and Diagnostic Tests
PD-L1 PD-L1 Second- line
Diagnostic Binding Criteria for
Drug Antibody Clone Domain Platform PD-L1 Positivity
Nivolumab 28-8 (rabbit) Extracellular Link 48 ≥1% tumor cells
(Bristol-Myers Squibb) Autostainer
Pembrolizumab 22C3 (mouse) Extracellular Link 48 ≥50% tumor cells
(Merck) Autostainer
Atezolizumab SP142 (rabbit) Cytoplasmica BenchMark Tumor cells and/or tumor-
(Genentech/Roche) ULTRA infiltrating immune cells
Durvalumab SP263 (rabbit) Cytoplasmica,b BenchMark ≥25% tumor cells

(AstraZeneca/
MedImmune)
Avelumab 73-10 unknown Dako assay ≥1% tumor cells
(Pfizer/Merck Serono)
a
Rebelatto 2016.
b
Epitope of AA249-290 membrane and intracytoplasmic. Antibody clones 28-8, 22C3, and 73-10, and Link 48 Autostainer are
products of Agilent Technologies/Dako. Antibody clones SP142 and SP263 and BenchMark and BenchMark ULTRA are products
of Ventana.
Postanalytic Phase
General
The postanalytic phase starts with microscopic evaluation of the stained slide(s). The inten-
sity of the staining is dependent on the enhancement (detection) system used (Figure 1).
The assessment of staining intensity is unavoidably subjective to some extent, but variation
in interpretation may be reduced with the following approach. The use of successive micro-
scope objective lenses with inherent related spatial resolution is a physical aid to establishing
the intensity level, as first applied to 20

HER2 testing (Rüschoff 2012). This Tyramide amplification
approach may lead to more unifor-

Polymer
Intensity (A.U.)
mity in staining intensity scoring.
Strong staining (3+) is clearly visible 10 Direct
with use of a x2 or x4 microscope labeling
objective lens, moderate staining
(2+) requires a x10 or x20 objective
lens to be clearly seen, and weak 0
1 10 80
staining (1+) can be seen only with –/+ Epitope Concentration
a x40 objective lens. The classic –/ + ++ +++
histo-score (H-score) is derived by
Figure 1. Relation between epitope concentration and signal enhance-
multiplying the percentage of tumor ment in immunohistochemistry (IHC). AU = arbitrary unit. Modified with
cells that stain positively by differ- permission from Prinsen CF, Klaassen CH, Thunnissen FB. Microarray as
ent intensities (0, 1, 2, or 3), yielding a model for quantitative visualization chemistry. Appl Immunohistochem
Mol Morphol. 2003;11(2):168-173.
a range of 0 to 300 for a total score.
This approach takes greater account of the heterogeneity of the staining. Interestingly,
with additional tyramide enhancement (staining amplification), the difference in epitope
concentration between a negative- and a strong-positive staining intensity is reduced to the
extent that the staining, and therefore scoring, is either “negative” or “positive,” meaning
simply “absent” or “present” (Figure 1). The strong signal enhancement may have important
consequences. A strong staining enhancement system can, in some cases, lead to a posi-
tive test result; whereas a test, on the same tissue, with weaker signal enhancement may
be negative. This factor is crucial, as it implies that once a test is clinically validated, only
tests with equal test performance can be used, otherwise the predictive performance of the
assays used in the phase III trials will not be realized.
Recently the term “immunohistochemistry critical assay performance control” (iCAPC)
was introduced. iCAPCs are external positive controls. iCAPCs monitor the overall system
performance but, like any other
external positive control, they Maximally
positive Optimal Suboptimal
do not fully inform about the IHC procedure IHC procedure
results with individual patient’s
Intensity (A.U.)
samples because final results also

substantially depend on preana-
lytic variables that are unique to
patients’ samples. The optimal IHC
positive control has an intensity Negative
performance at or above the low 0 A B High
limit of detection and is defined Epitope Concentration
by an observed positive reaction Figure 2. The relation between epitope concentration and intensity for an
(staining) in a tissue/cellular ele- optimal and a suboptimal IHC staining procedure. In the optimal procedure,
the staining intensity at epitope concentration (A) will be weak positive but
ment that is known to express low
will be negative in the suboptimal procedure. Also, when a control has a
levels of the evaluated marker high epitope concentration (B) there will not be a difference between the
(Figure 2) (Torlakovics 2015). two procedures. AU = arbitrary unit.
The use of an internal positive control tissue is frequently perceived as adequate, and an
external control may not be needed. However, the advice for most stains is to maintain the
external control tissue because not every section will contain the tissue element in ques-
tion, and internal positive controls usually have a relatively high epitope concentration.
Therefore, internal positive control tissues rarely provide information if the appropriate
sensitivity of the protocol was achieved. This gap is filled by the external positive control
tissue with low epitope concentration, tissue placed on the same slide as the tissue to be
tested (Figure 2).
Standardization of both positive and negative controls is needed for diagnostic and predic-
tive IHC. In general, the use of IHC-negative controls, regardless of type, is not standardized
from a global point of view, although this use is well established. As such, the relevance and
applicability of negative controls continues to challenge both pathologists and laboratory
budgets. Despite the clear theoretical notion that appropriate controls serve to demonstrate
the sensitivity and specificity of the IHC test, it remains unclear which types of positive and
negative controls are applicable and/or useful in daily clinical practice (Torlakovic 2014,
Torlakovic 2015).
PD-L1 Validation Table 3. Comparison of Analytic and Clinical Validation in

Development of histologic criteria Immunohistochemistry
and interpretation of the PD-L1 assay Validation
depends on the application. The crite- Factors Analytic Clinical a
ria used are associated with differences
Preanalytic steps within limits Yes Yes
in response rates related to the use of
specific treatment. This understanding Analytic steps robust Yes Yes
can only be achieved by comparison Minimum number of positive 10 10
of a study group, for which treatment cases for validationb
with the intended drug was performed, Minimum number of negative 10 10
and data on outcome and IHC bio- cases for validationb
marker (preferably graded) are known. Minimum number covering lin- – 20
Moreover, the predictive procedure, ear dynamic rangec
including preanalytic, analytic, and Requires treatment outcome in No Yes

interpretation phases must be robust study group for a certain druga 3
to maintain the predictive value. The Test must be constant in time to No Yes
College of American Pathologists maintain predictive value
has established principles of analytic a
Clinical validation may be obtained by clinical samples with treat-
validation, and principles of clinical ment response data used for threshold determination in a certain test
with its linear dynamic range (Rebelatto 2016). Once a threshold is set
validation have also been set forth in a certain test, demonstration of equivalence of an alternative test in
(Table 3) (Fitzgibbons 2014). This need the region of the linear dynamic range is needed for clinical validation.
b
for absolutely reliable and consistent The minimum number of positive and negative cases are according
to guidelines of the College of American Pathologists (Fitzgibbons
biomarker tests, with proven, equiva- 2014). These guidelines are not evidence-based but rather represent
lent clinical, predictive, and technical expert opinion, and every laboratory director can decide to validate
using fewer or more samples.
performance, raises the important
c
The samples covering the linear dynamic range (see Figure 1) are the
issue of whether to use LDTs or com- most relevant for clinical validation because the threshold for positiv-
mercial kit assays. ity is within this group. This number is based on expert opinion.
In establishing the clinical validity of an LDT, the IHC should be performed in the same
way as the corresponding clinically validated commercial test. For example, the antibody
should be titrated; the incubation time should be varied to obtain the same signal; the effect
of the signal enhancement system should be equal; the PD-L1 positive areas should be the
same in serial sections; and the positive areas should be the same for approximately 10 PD-L1
negative samples, 10 PD-L1 positive samples, and 20 samples covering the linear dynamic
range of the clinically validated PD-L1 IHC test. Conceptually, samples that are positive in
a clinically validated test and become negative when the primary antibody is diluted to 25%
are within the linear dynamic range. Samples that remain positive are uninformative for
comparison of different clinically validated assays on the same protein (eg, PD-L1). Samples
that are negative using a clinically validated test and turn positive when the concentration
of the primary is increased fourfold are also within the linear dynamic range. However,
if such a sample remains negative, it is not informative for comparison of different clini-
cally validated assays on the same protein. As outcome measures, the intraobserver and
interobserver variations should be kept within reasonable ranges. For clinical validation,
the laboratory should be certified and follow standard operating procedures (including vali-
dation reports) for commercial kits and LDTs so as to maintain robust testing in time. The
laboratory also should regularly and successfully participate in external quality-assessment
schemes. If any of these requirements is not fulfilled, predictive performance of the assay
cannot be guaranteed.
PD-L1 Interpretation “Histology”

PD-L1 expression may be present on dendritic cells, macrophages, mast cells, and T- and
B-lymphocytes, as well as on endothelial and tumor cells (Yu 2016). PD-L1 has two small
hydrophilic regions for binding sites of IHC detection antibodies (Patel 2015). The biologic
consequences of B7-H1 expression depends on cell membrane localization because it is pre-
sumed that B7-H1 is functional only when it ligates into the counter-receptor (Sznol 2013).
Before examining a patient specimen for PD-L1 staining, it is important to first exam-
ine the hematoxylin and eosin (H&E) stain
to assess preservation and staining quality.
Then, the external positive and negative
control tissue slides should be examined. If
any staining of the PD-L1 IHC external con-
trol slide is not satisfactory, all results with
the patient specimens should be considered
invalid. For PD-L1, external positive control
FFPE cell blocks of human tonsils (Figure 3)
and/or cell lines with variable PD-L1 expres-
sion (such as those available from Histocyte
Laboratories, Tyne and Wear, United Figure 3. Tonsil staining with the programmed cell
Kingdom) may be used to set up the staining death-ligand 1 (PD-L1) IHC 22C3 pharmDx (Agilent
conditions. Care should be taken when using Technologies/Dako), showing membrane staining in the
crypt epithelium (left) and scattered membranous staining
only cell lines with high epitope concentra- of macrophages within the lymphoid follicles (right upper)
tions because the optimal staining control (x20 magnification).
has weak epitope concentration, allowing detection of minor deviations in the staining
protocol.
Examination of the Patient Slide

When patient slides are examined, necrotic or degenerated malignant cells should be
excluded from evaluation (Figure 4). A minimum number of tumor cells defined by the
assays should be present in the PD-L1-stained patient slide to determine the proportion of
stained tumor cells. The minimum number of tumor cells is defined by the manufacturer
of the assay; the minimum is 50 cells for the SP142 assay (Ventana) and is 100 cells for the
28-8 and 22C3 pharmDx assays (Agilent Technologies/Dako). Immune cells, such as infil-
trating lymphocytes or macrophages, may also serve as PD-L1 positive internal controls.
Any background staining greater than 1+ staining intensity is unacceptable (Figure 5).
Figure 4. Pulmonary adenocarcinoma stained with Figure 5. Non-small cell lung cancer (NSCLC) sample
the programmed cell death-ligand 1 (PD-L1) IHC 22C3 stained with the anti-programmed cell death-ligand 1
pharmDx (Agilent Technologies/Dako), showing necrotic (PD-L1) 22C3 antibody, demonstrating unacceptable
areas (x40 magnification). nonspecific background staining (more than 1+) (x20
magnification).
Examination of the Negative Non-small Cell Lung Cancer (NSCLC) Tissue

The negative NSCLC tissue is examined to ascertain that there is no unintended staining.
Any background staining should be of less than 1+ staining intensity. If plasma membrane
staining of malignant cells occurs in the negative control tissue, all results with the patient
specimens should be questioned, and the negative control specimen should be replaced by
another negative control in a repeat analysis. Even if the slide appears to have no staining
reactions, the slides should be viewed at high magnification for confirmation so as not to
overlook weak membrane staining.
PD-L1 Interpretation “Cytology/Cell Blocks”

Although at least 30% to 40% of all patients with advanced NSCLC are diagnosed by cytology
alone, the use of cytology samples for determination of PD-L1 is not advocated because none
of the assays are validated for this purpose. However, because immunostaining of tumor
cells on cytology samples is standard diagnostic practice in many institutions (Fischer 2014,
Savic 2015), PD-L1 staining and quantitation may be feasible in principle, provided that
appropriate protocols and quality-control measures are in place. In case of alcohol fixation
the PD-L1 IHC protocol may require adjustment. Actually, one author (L.B.) demonstrated
promising preliminary results comparing PD-L1 tumor cell staining on ethanol-fixed and
Papanicolaou-stained smears with matched histologies (Figure 6).
In cytology samples, quantitation of PD-L1-positive immune cells using the SP142 assay
will likely be more challenging. The lack of tissue architecture precludes distinction of the
relevant immune cells between the tumor cells and at the epithelial-stromal interface from
immune cells that are outside of the tumor boundaries and are considered as being irrelevant
Figure 6. Programmed cell death-ligand 1 (PD-L1) immunohistochemistry (IHC) analysis of cytologic

specimens. All stained for PD-L1 with 28-8 antibody. (A) A biopsy tumor sample with membranous
and cytoplasmic PD-L1-positive tumor cells. The membranous positive fraction is lower (+/- 10%)
than those with only cytoplasmic staining (+/- 60%). (B) Tumor cells strongly 100% PD-L1-positive
(membranous and cytoplasmic) with also macrophage staining in stroma. (C) Tumor cells are PD-L1-
negative, while necrotic cells are cytoplasmic-positive (middle right). (D) Pre-existing mucus gland-
negative, while stromal immune cells are PD-L1-positive. (E) Tumor cells are negative for PD-L1, but
lymphocytes and macrophages are positive. Note occasional membranous positive macrophage in
between tumor cells. (F) Peripheral lung biopsy with collapsed tissue. Intra-avleolar macrophages
are PD-L1-membranous positive.
for PD-L1 scoring. Moreover, pre-existing lymphocytes in a fine-needle aspirate of a lymph

node are a major confounding factor. Although peer-reviewed published literature is not
available at the time of publication, emerging data from cell blocks and matched histologic
specimens suggest that cytologic material is as good as histologic material for PD-L1 IHC
tumor cell analysis (Skov 2016).
Definition of Positive PD-L1 Staining in NSCLC

Not all definitions for PD-L1 positivity are the same for the five assays. In four of them–28-
8, 22C3, SP263, and 73-10–positive PD-L1 staining is defined as complete circumferential
or partial linear plasma membrane staining of tumor cells at any intensity. Cytoplasmic
staining in tumor cells is not considered positive for scoring purposes. Nonmalignant cells
and immune cells, such as infiltrating lymphocytes or macrophages, and necrosis may also
stain positively for PD-L1; however, these cells should not be included in the scoring for
the determination of PD-L1 positivity of tumor cells.
In the assay using the SP142 antibody clone, the PD-L1-positive immune cells, as well
as the tumor cells, are considered in the criteria of positive PD-L1 staining (See Chapter 6).
To a certain extent, this hampers the attempt to establish one PD-L1 IHC test that will be
equivalent to all clinically validated tests and will possibly lead to the SP142 test having to
be retained for use with atezolizumab.
PD-L1 Scoring
As a general scoring procedure, the tumor areas of the entire specimen are first carefully
examined at x4 objective magnification. Well-preserved and well-stained areas of the speci-
men should be used to evaluate PD-L1 staining. Next, the specimen is examined at x10 to
x40 objective magnification, and viable tumor cells exhibiting complete circumferential or
partial linear plasma membrane staining at any intensity are scored. Cytoplasmic staining
is excluded from scoring. With the SP142 assay, immune cells are part of the scoring algo-
rithm only when atezolizumab therapy is being considered. Normal and necrotic cells are
always excluded from scoring.
As the expression of PD-L1 is a continuous variable, any scoring around any of the
relevant thresholds will inevitably be subject to some interobserver variability that could
lead to an alternative clinical decision regarding therapy. A recent ring trial provides some
information about this variability in a group of 15 patients with NSCLC and nine observers
(Table 4) (Scheel 2016; Cooper 2016). Overall, inaccuracy of scoring due to interobserver
discordance is less than 10%. (Further details will be discussed in Chapter 10.)
Table 4. Interobserver Variation in Scoring for a Highly Selected Set of CasesAntibody
Antibody
Study Threshold 28-8 22C3 SP142 SP263
Scheel 20161 1%/50% 2.8%/5.2% 7.4%/8.3% 9.6%/8.5% 6.3%/7.4%
Cooper 20162 1%/50% 15.8%/18.1%
Note that the variation in clinical practice is not known yet and needs to be determined on a consecutive series of
cases. 2) Mean of 2,700 pairwise comparisons, 10 pathologists, 108 (highly selected set of samples). 1) These per-
centages are obtained using a selection of 15 NSCLC resection specimens after some training by nine pathologists.
Modified from Scheel et al. Mod Pathol. 2016;29(10):1165-1172.
PD-L1 Heterogeneity
As PD-L1 expression is often patchy, it is recommended for larger samples to divide the
tumor on the slides being assessed into areas of equal amount of tumor at low magnifica-
tion and evaluate each area for PD-L1 positivity. The average percentage of positivity from
all areas is the overall percentage tumor proportion score for PD-L1 positivity (Figure 7).
PD-L1 expression at the tumor–stroma interface may be enhanced because of an immune
activation response. This interface aspect contributes to heterogeneity of tumor cell PD-L1
expression, and smaller tumor biopsy samples may be missing the pertinent tumor–immune
interface, leading to a possible difference in PD-L1 expression outcome between biopsy and
surgical specimens due to sampling. The concordance between biopsy and resection was 92%
in one study (Kitazono 2015) but was lower (52%) in another study, with underestimations
in the biopsy specimens (Illie 2016). However, the latter study was unclear in reporting the
pre-analytic variation. This aspect clearly needs more study.
As in daily practice, more biopsy than resection specimens will be stained for PD-L1,
because immune checkpoint therapy is currently indicated only for patients with advanced
NSCLC. Thus, an inaccuracy in PD-L1 testing due to sampling of heterogeneous tumors is
unavoidable. The fact that approximately 10% of NSCLC tumors respond to PD-L1/PD-1
inhibitors despite absence of PD-L1 expression may be partly explained by false-negative
results on biopsy specimens of PD-L1-positive tumors with heterogeneity.
Interpretation Pitfalls
As for all IHC stainings, artifacts may be due to nonspecific background that occurs because
of improper drying, improper deparaffinization, or incomplete rinsing of the slides (Figure
5); edge artifacts due to drying of the tissue prior to fixation or during the staining procedure
(Figure 8); crush artifacts (Figure 9); necrosis; or poor fixation (Figure 10). PD-L1-positive
lymphocytes and histiocytes may lie in between PD-L1 negative tumor cells and be inter-
preted as positive. Alveolar macrophages (Figure 11) may have membranous staining and
may be used as an internal positive control, but they can be falsely interpreted as positive
tumor cells when they are close to or adjacent to PD-L1-negative tumor cells. The nuclear/
cytoplasmic ratio, a thin nuclear membrane of the macrophages, and their context in the
sample may be helpful clues. In difficult cases, an IHC analysis using a macrophage marker
can be helpful. Cytoplasmic staining of tumor cells may be granular but this staining should
not be considered positive.
General Reporting Practices

Although practices vary, it is strongly recommended that pathologists interpret the results
of all positive controls as an integral part of interpreting and reporting the results of IHC
staining. In addition, predictive markers should never be evaluated in the absence of ref-
erence controls (Torlakovic 2014). Assuming that these recommendations are part of the
standard operating procedure and that controls are interpreted as adequate, these details do
not need to be written in the patient report. In practice, the name of the diagnostic kit and
the diagnostic criteria used should be reported. In cases where PD-L1 staining is absent in
the tumor, the adequacy of the PD-L1 control section staining should be mentioned. Specific
reporting details for each assay are detailed in Chapters 4 through 8. Because therapeutic
1 2
3 4
B C
D E
F
Figure 7. Example of larger specimen with heterogenous
staining for PD-L1 (28-8). (A) Shows 4 adjacent fields
(1-4, 5x objective), which are individually represented
in B, C, D, and E (10x objective), respectively. (B-E) For
an estimation of PD-L1 tumor proportion score (TPS),
the TPS of the four individual fields is averaged. (F) If
needed, a detailed impression of membranous PD-L1
staining may be examined at higher magnification (20x
objective). In this case, the TPS of B-E is 90%, 40%, 80%,
and 30%, respectively. This leads to a TPS for A of 60%
[=(90+40+80+30)/4]. Note that the necrotic areas also
stain cytoplasmic with PD-L1. Necrotic areas are by
definition excluded for PD-L1 TPS estimation.
Figure 8. Non-small cell lung cancer (NSCLC) sample Figure 9. Non-small cell lung cancer (NSCLC) stained
stained with the 22C3 antibody, showing edge artifact with the 22C3 antibody, showing crush artifact (x10
in staining (x4 magnification). Edge staining should be magnification).
excluded from the scoring.
Figure 10. Pulmonary adenocarcinoma with Figure 11. Intra-alveolar macrophages containing
poor tissue fixation, showing an ambiguous pro- anthracotic pigments stained with the 22C3 antibody
grammed cell death-ligand 1 staining pattern (x20 (x40 magnification). These macrophages should be
magnification). excluded from the scoring.
response of immune checkpoint inhibitors is reported to be in proportion to the extent of

PD-L1 reactivity, reporting of the extent of positive tumor cells, at least in 10% increments,
is recommended. If the immunotherapeutic agent to be used is known at the time of test-
ing, the results can be reported in terms of broader categories (eg, <1%, 1% to 49%, >50%),
appropriate for the drug to be used.
Conclusion
PD-L1 IHC is a biomarker with predictive value for immunotherapy. Pathology laborato-
ries should use at least one validated test that must be affordable, given the expected high
volume. The requirements of such a test and its usage, in general terms, have been described
in this chapter. The reader is referred to the assay-specific chapters in this Atlas for more
information.
PD-L1 28-8 pharmDx Assay
By Sylvie Lantuéjoul and Erik Thunnissen 4
The PD-L1 IHC 28-8 pharmDx (Agilent Technologies/Dako) is a laboratory test that measures
programmed cell death ligand-1 (PD-L1) expression in formalin-fixed, paraffin-embedded
(FFPE) tissue samples on the Autostainer Link 48 platform (Agilent Technologies/Dako)
(Phillips 2015). This assay is considered a complementary diagnostic tool for the treatment of
patients with advanced non-small cell lung cancer (NSCLC) with nivolumab (see Chapter 10 for
details). However, PD-L1 testing is not required as a selection biomarker to treat patients with
either squamous or non-squamous lung cancer with nivolumab (Brahmer 2015, Borghaei 2015).
Nivolumab is a human immunoglobulin G4 (IgG4) monoclonal antibody that binds to the
programmed cell death protein-1 (PD-1) receptor and blocks its interaction with PD-L1 and
PD-L2, restoring the antitumor immune response (Wang 2014). It has been approved by the
US Food and Drug Administration (FDA) and the European Commission to treat patients
with advanced (metastatic) NSCLC whose disease progressed during or after platinum-based
chemotherapy (second-line therapy), because of enhanced survival.
Antibody Characteristics and Immunostaining Conditions

Clone 28-8 (ab205921; Abcam) is an IgG4 isotype rabbit monoclonal anti–(human) PD-L1
antibody. Its immunogen is a recombinant full-length protein that corresponds to the extra-
cellular domain (Phe19-Thr239) of human PD-L1 (Phillips 2015). Clone 28-8 detects PD-L1
protein on FFPE specimens. Human tonsil is a recommended positive tissue control, with
highest expression of PD-L1 in the crypt epithelium, macrophages homing the germinal
centers, and interfollicular mononuclear leukocytes. Cell lines, such as B-CPAP (a human
papillary thyroid cancer cell line with high expression), ES-2 (an ovarian clear cell carcinoma
cell line with intermediate expression), and HCC70 (a ductal carcinoma cell line with low
expression) can also be used as external positive controls. This PD-L1 primary antibody
showed no cross-reactivity for PD-L2 exogenously expressed in Chinese hamster ovary
(CHO) cells (an epithelial cell line).
The PD-L1 IHC 28-8 pharmDx is an FDA-approved and European Conformity (CE) In
Vitro Device (IVD)-marked, qualitative, IHC assay that contains ready-to-use optimized
reagents and includes a protocol to complete an IHC staining procedure on the Autostainer
Link 48. Positive and negative control slides are provided (pelleted, FFPE of PD-L1–positive
NCI-H226 squamous cell carcinoma/mesothelioma cell lines and PD-L1–negative MCF-7
breast adenocarcinoma cell lines). The signal-enhancement system is the EnVision FLEX
visualization system. The staining procedure contains a series of steps: the epitope-retrieval
solution EnVision FLEX Target Retrieval Solution, low pH is used, followed by a peroxidase-
blocking reagent. Then, in parallel, the primary monoclonal rabbit anti–PD-L1 antibody
clone 28-8 is incubated on the intended positive-control slide, and the negative control
reagent is incubated on the intended negative-control slide. Both slides are treated similarly
throughout the following steps: washing; incubation with a rabbit linker peptide; use of a
visualization reagent containing a polymer labeled with horseradish peroxidase enzymes;
use of a visualization reagent consisting of secondary antibody molecules and horserad-
ish peroxidase molecules, coupled to a dextran polymer backbone; addition of chromogen
3, 3’ -diaminobenzidine in a (usually clear) buffer solution and enzymatic conversion result-
ing in precipitation of a visible (brown) reaction product at the site of the antigen; and
addition of a chromogen-enhancement reagent to convert the reaction product to a dark
brown color. The section is then counterstained, covered with a mounting fluid that has a
similar index of refraction as glass (1.5), and cover-slipped. All of the required steps and
incubation times for staining are pre-programmed in the DakoLink software.
According to the manufacturer, the materials provided in each 28-8 pharmDx kit are suf-
ficient for 50 tests (50 slides with the primary antibody and 50 slides with the corresponding
negative control reagent) and for a maximum of 15 individual staining runs. The kit also
provides additional primary antibody to stain 15 cell line control slides. The number of tests
is based on the use of 2 x 150 μL per slide of each reagent except for 3, 3-diaminobenzidine
positive and Envision FLEX Target Retrieval Solution.
Evaluation of Staining and Reporting

A minimum of 100 viable tumor cells are required to determine the percentage of stained
tumor cells per slide for PD-L1 assessment. Nonmalignant and immune cells (eg, infiltrat-
ing lymphocytes or macrophages) may also stain for PD-L1 expression. The manufacturer
recommends staining control slides (containing FFPE positive and negative cell lines- see
above), a slide with the negative control reagent for each patient case, as well as the use of
laboratory-supplied positive and negative control tissue slides.
PD-L1 staining is defined as complete circumferential or partial linear plasma mem-
brane staining at any intensity. Cytoplasmic staining, if present, is not considered positive
for scoring purposes. The percentage of viable tumor cells exhibiting positive-membrane
staining at any intensity in the entire specimen may be reported as less than 1%, 1% to less
than 5%, 5% to less than 10%, and 10% or greater. Because tumors may heterogeneously
express PD-L1 (Figures 1 and 2), the specimen should be divided in areas of equal propor-
tion of positive cells at low magnification, with an evaluation of each area for percentage of
PD-L1 positivity. The PD-L1 positivity percentages from each area are then added together
and divided by the total number of areas, to reach the final percent of PD-L1 positivity (see
Chapter 3 for details).
PD-L1 28-8 PHARMDX ASSAY 51
Figure 1. Programmed cell death ligand-1 (PD-L1) Figure 2. Programmed cell death ligand-1 (PD-L1) immu-
immunohistochemistry (IHC) using a clone 28-8 anti- nohistochemistry (IHC) using a clone 28-8 antibody for
body for squamous cell carcinoma. Strong membrane adenocarcinoma. Heterogeneity of staining with variable
staining of all tumor cells is shown using immunoper- intensities is showing using immunoperoxidase (x200
oxidase (x200 magnification). magnification).
Several details are suggested when reporting Box 1. Suggested Information to Include
results with the PD-L1 IHC 28-8 pharmDx assay When Reporting Results from the PD-L1 IHC
28-8 pharmDx Assay
(Box 1).
General Information
• Positive control results (Pass/Fail)
Interpretation Pitfalls • Negative control results (Pass/Fail)
Artifacts may be due to nonspecific background (ie, • Adequate tumor cells (at least 100 cells)
improper drying of the slides, improper deparaf- are present (Yes/No)
finization, or incomplete rinsing), edge artifacts • Tumor Proportion Score:
PD-L1 Expression
(ie, drying of the tissue prior to fixation or during • < 1% ___
the staining procedure), crush artifacts, necrosis, • ≥1% ___
or poor fixation. • ≥ 5% ___
Staining may be interpreted as a false-positive • ≥ 10% ___
result in samples where positive tumor-infiltrat- Optional Information
ing lymphocytes and macrophages are intimately • Presence/amount of tumor-associated
admixed with tumor cells. Granular cytoplasmic immune cells
staining in the absence of membranous staining • PD-L1 positivity in increments of 10%
• Other comments to the clinician
should not be interpreted as positive. Alveolar mac-
rophages frequently show membranous staining
and may be mistaken for tumor cells by an inexperienced reader. The nuclear/cytoplasmic
ratio and thin nuclear membrane of the macrophages may be helpful clues. In contrast to
viable tumor cells, necrotic tumor cells often show cytoplasmic-only staining.
Several factors may lead to false-positive interpretations. It is possible for PD-L1–positive
lymphocytes/histiocytes to lay in between PD-L1–negative tumor cells, but the overall speci-
men could be interpreted as positive (Figure 3). Cytoplasmic staining of tumor cells might
be granular but not membranous and, therefore, interpreted as positive. It is imperative to
be careful when judging the context of results (Figure 4). In addition, necrotic cells might be
interpreted as positive, but these usually have cytoplasmic distribution (not membranous,
Figure 5).
Figure 3. Programmed cell death ligand-1 (PD-L1) immu-

nohistochemistry (IHC) using a clone 28-8 antibody for
adenocarcinoma. Positive immune cells infiltrating tumor
lobules with no positive malignant cells are shown using Figure 4. Programmed cell death ligand-1 (PD-L1) immu-
immunoperoxidase (x200 magnification). nohistochemistry (IHC) using a clone 28-8 antibody for
squamous cell carcinoma. Alveolar macrophages (star)
can be seen mixed with tumor cells (arrow) using immu-
noperoxidase (x200 magnification). Macrophages are
Predictive Significance in Lung Cancer positively stained but weaker than tumor cells.
Clinical utility of the PD-L1 IHC 28-8 pharmDx
assay was evaluated in two phase III, random-
ized, open-label studies of nivolumab compared
with docetaxel in patients older than 18 years
who had advanced or metastatic squamous
(Brahmer 2015) and non-squamous NSCLC
(Borghaei 2015) for whom treatment with a
platinum-based chemotherapy doublet failed
(Table 1, and see Chapter 2 for details). The
results of these studies were used to gain FDA
approval of nivolumab in the second-line treat-
ment of patients with advanced-stage squamous
and non-squamous cell lung cancer.
In Checkmate 017, primary and secondary
endpoints were reached, with a 9.2-month over-
all survival for patients treated with nivolumab Figure 5. Programmed cell death ligand-1 (PD-L1) immu-
nohistochemistry (IHC) using a clone 28-8 antibody for
versus 6 months for patients treated with squamous cell carcinoma. An area of necrosis (star) is
docetaxel. The median progression-free survival visible, as is membrane staining of surrounding tumor
was 3.5 months and 2.8 months, respectively. cells (arrow) (immunoperoxidase; x200 magnification).
The risk of death was 41% lower with nivolumab
than with docetaxel (HR 0.59; 95% CI [0.44-0.79]; p < 0.001). At 1 year, the overall survival
rates were 42% (95% CI: 34-50) for patients treated with nivolumab and 24% (95% CI:
17-31) for docetaxel. The response rates were 20% with nivolumab and 9% with docetaxel
(p = 0.008). However, PD-L1 positivity at any cutoff was neither significantly prognostic
nor predictive in squamous histology, but the size of the cohort was too small. There was a
trend toward improved overall survival and overall response rate for patients with PD-L1
expression of 1% or greater, which could possibly indicate a long-term benefit (Brahmer
2015). (Figure 6 illustrates a dramatic response to nivolumab for a patient who received
PD-L1 28-8 PHARMDX ASSAY 53
Table 1. Predictive Significance of Results from the PD-L1 IHC 28-8 pharmDx Assay in Lung Cancer
No. (%) of Pts. with PD-L1
Expression
No. of Pts.
Evaluable
First Author (Year), Tumor Stage, for PD-L1
Trial Histology Expression ≥ 1% ≥ 5% ≥ 10%
Brahmer (2015), IIIB-IV squamous cell 225 119 (53) 81 (36) 69 (31)
CheckMate 017 NSCLC
Borghaei (2015), IIIB-IV nonsquamous 455 246 (54) 181 (40) 165 (36)
CheckMate 057 cell NSCLC
Rivzi (2016), IIIB-IV NSCLC 44 23 (52) NR NR
CheckMate 012
(combination with
chemotherapy)
Gettinger (2016), IIIB-IV Squamous and 46 32 (70) NR NR
CheckMate 012 nonsquamous NSCLC
(monotherapy)
Rivzi (2015), IIIB-IV squamous 76 NR 25 (33) NR
CheckMate 063 NSCLC
Antonia (2016), SCLC (all stages) 213 24 (11%) 7 (3%) NR
CheckMate 032
* Number of cases evaluable for programmed cell death ligand-1 (PD-L1) expression.
NSCLC = non-small cell lung cancer, NR = not reported, SCLC = small cell lung cancer.
Figure 6. A 68-year-old male smoker with cT2aN1M1b (OSS, BRA) non-small cell lung cancer (adenocarcinoma). Because EGFR,
ALK, and KRAS were wild type, the patient was treated with nivolumab after chemotherapy, which led to dramatic shrinkage
of the tumor. Disease on computed tomography images before (top left) and after (top right) treatment with nivolumab.
Staining with the PD-L1 IHC 28-8 pharmDx assay (bottom row, middle) showed diffuse positive reaction of the tumor cells.
prior chemotherapy and was subsequently treated with nivolumab.) Eventually, because a
significant proportion of patients with PD-L1–negative tumors benefitted from treatment
with nivolumab, the FDA did not require PD-L1 testing before treatment. The FDA has
approved the PD-L1 IHC 28-8 pharmDx assay, however, to detect PD-L1 expression levels
and to help physicians determine which patients may benefit most from treatment with
nivolumab. In Checkmate 057, the median overall survival was 12.2 months (95% CI: 9.7-
15.0) for patients treated with nivolumab and 9.4 months (95% CI: 8.1-10.7) for docetaxel.
At 1 year, the overall survival rate was 51% (95% CI: 45-56) with nivolumab versus 39%
(95% CI: 33-45) with docetaxel. The response rate was 19% with nivolumab versus 12% with
docetaxel (p = 0.02). Regarding PD-L1 expression, nivolumab was associated with greater
efficacy than docetaxel in subgroups, which were defined according to prespecified levels
of tumor-membrane expression of PD-L1 (1% or greater, 5% or greater, and 10% or greater).
The median overall survival for patients in these subgroups was 17.1, 18.2, and 19.4 months,
respectively, with nivolumab compared with 9.0, 8.1, and 8.0 months, respectively, with
docetaxel (Borghaei 2015).
Nivolumab was investigated as monotherapy for first-line management of advanced
NSCLC in the phase I multicohort CheckMate 012 trial. Overall response rates were 28%
for patients with any degree of tumor PD-L1 expression and 14% for patients with PD-L1–
negative tumors. The median progression-free survival was 3.6 months, the median overall
survival was 19.4 months, and the 1-year and 18-month overall survival rates, respectively,
were 73% (95% CI, 59-83) and 57% (95% CI, 42-70) (Rizvi 2016, Gettinger 2016).
Recently, the CheckMate 026 trial, a phase III, open-label, randomized study of nivolumab
as monotherapy versus the investigators’ choice of chemotherapy for patients with advanced
NSCLC did not meet its primary endpoint of progression-free survival in patients with
previously untreated advanced NSCLC whose tumors expressed PD-L1 at 5% or greater
(Socinski 2016, see Chapter 2 for details).
Conclusion
The PD-L1 IHC 28-8 pharmDx assay is a complementary diagnostic tool for the management
of non-squamous NSCLC with PD-L1 expression of 1% or greater using nivolumab. This test
has been validated for FFPE tissue samples and is used on the Autostainer Link 48 platform.
PD-L1 IHC testing is not required to treat patients with squamous NSCLC with nivolumab.
PD-L1 22C3 pharmDx Assay
By Teh-Ying Chou, Wendy A. Cooper, and Keith M. Kerr 5
Introduction of the Platform
PD-L1 IHC 22C3 pharmDx assay (Dako, Glostrup, Denmark) is an in vitro diagnostic
(IVD) immunohistochemistry (IHC) assay for detection of PD-L1 protein expression in
non-small cell lung cancer (NSCLC) tissue (Dako 2016). This assay is performed on the
Dako Autostainer Link 48 platform with an automated staining protocol using a mouse
monoclonal anti-PD-L1 antibody, clone 22C3. The assay is indicated as an aid in identifying
advanced-stage NSCLC patients who would be eligible for treatment with pembrolizumab
(KEYTRUDA®, Merck Sharp & Dohme Corp.), a humanized monoclonal IgG4 kappa isotype
antibody against PD-1. In October 2015, the PD-L1 IHC 22C3 pharmDx assay was approved
by the US Food and Drug Administration (FDA) as a companion diagnostic test for treat-
ment with pembrolizumab in patients with advanced (metastatic) NSCLC (FDA, 2016). This
assay assesses PD-L1 protein expression by evaluating “tumor proportion score” (TPS),
which is the percentage of viable tumor cells showing either partial or complete membrane
staining (Dako 2016). Increased PD-L1 expression (higher TPS) is generally associated
with higher objective response rate (ORR) and favorable outcome in patients treated with
pembrolizumab (Baas 2016).
Antibody Characteristics and Immunostaining Conditions

PD-L1 IHC 22C3 pharmDx uses a mouse monoclonal antibody clone 22C3 that recognizes
the extracellular domain of PD-L1 to assess the PD-L1 expression in formalin-fixed, paraffin-
embedded (FFPE) tissue with use of IHC. The IHC staining procedure is performed on Dako
Autostainer Link 48 platform with a validated staining protocol. Briefly, slides are baked at
60°C for 30 minutes and then subjected to deparaffinization, rehydration, and target retrieval
on the Dako PT Link Pre-Treatment Module (Dako No. PT100) using the Dako EnVision
FLEX Target Retrieval Solution, Low pH. The following staining procedure is performed
on Dako Automated Link 48 platform. First, the slides are incubated with anti-PD-L1 22C3
antibody or a negative control reagent (mouse IgG isotype) for 20 minutes. Subsequently,
Dako EnVision FLEX+ Polymer Reagents, including a mouse linker, horseradish peroxidase
polymer, diaminobenzidine chromogen (DAB), and DAB enhancer are used for primary
antibody detection. The EnVision FLEX+ Wash Buffer is applied for washing between each
reaction step. After primary antibody detection, the slides are counterstained with hema-
toxylin and coverslipped. Staining results are interpreted with use of a light microscope.
Although the PD-L1 IHC 22C3 pharmDx assay is a companion diagnostic test that
is well standardized and validated, some unexpected issues, such as batch-to-batch varia-
tions of reagents and errors from automatic instruments, may occasionally occur (Dako
2016; Cree 2016). Laboratories are recommended to include a variety of controls along with
clinical cases for PD-L1 IHC testing to ensure the assay performance (Table 1). The PD-L1
IHC 22C3 pharmDx assay provides a control slide containing FFPE sections of two pelleted
cell lines: NCI-H266 (a NSCLC cell line with moderate expression of PD-L1) and MCF-7 (a
breast adenocarcinoma cell line with negative expression of PD-L1). A control slide should
be stained with anti-PD-L1 22C3 antibody in each staining run to assess the validity of
staining. In addition to the control slides supplied in the kit, inhouse tissue controls also
should be regularly performed since the differences in the preanalytical phase, such as time
to fixation, fixation time, and tissue processing, etc., may result in significant variations of
staining. NSCLC tissues showing areas with at least positive and negative expression of PD-L1
are ideally chosen as inhouse controls; however, human tonsil or placenta tissues processed
in the same manner as the patients’ samples may be used as an alternative positive control.
In addition, use of the controls that demonstrate expression results close to the decision-
making cutoff points is recommended to assess the performance more sensitively. Notably,
control sections should be cut at the same time as the patients’ sample. Long-term storage of
pre-cut control sections may result in reduction of the antigenicity and should be avoided.
Table 1. Clinical Case and Control Slides Used for PD-L1 IHC 22C3 pharmDx Assay
Type Primary Ab Used Purpose Duration
Positive control Anti-PD-L1 antibody For control of all steps of the Regularly performed
(in house tissue) assay from pre-analytical phase
to analytical phase
Negative control Anti-PD-L1 antibody For detection of unintended Regularly performed
(in-house tissue) antibody cross-reactivity
Control slide supplied Anti-PD-L1 antibody For control of staining Performed in each run
by the kit procedure (analytical phase)
Patient tissue slide Mouse IgG For examination of the Performed in each run
presence of non-specific
background staining

PD-L1 IHC should be evaluated and scored by a qualified pathologist under light microscope.
Before examining the patient specimen, evaluation of the quality of the controls is indis-
pensable. It is recommended to obtain three serial tissue sections to perform hematoxylin
and eosin (H&E), PD-L1, and negative control reagent stains. The H&E is assessed first and,
if acceptable, the remaining two immunohistochemical stains are subsequently performed.
Each PD-L1 IHC 22C3 pharmDx is configured with control cell line slides that should be
PD-L1 22C3 PHARMDX ASSAY 57
included in each IHC run. Both the control cell line slide and patient-tissue control slide
for non-specific background staining should be assessed with every IHC run (Dako 2016).
At least 100 viable tumor cells are required for a valid interpretation of PD-L1 staining, as
well as for evaluation of positive control and negative control reagent stains. Therefore, the
evaluation of serial sections from the same paraffin block of the patient specimen is impor-
tant. If the patient specimen sections harbor fewer than 100 viable tumor cells, a deeper
level of sections (if judged likely to be helpful) or another block of choice (if available) are
suggested to obtain a sufficient number of viable tumor cells.
Examination of the control cell line slide is essential for determining whether the reagents
are functioning properly. Each control cell line slide contains both positive and negative cell
pellets. If the staining of the control cell line slide is unsatisfactory, the result for the patient
specimen should be considered invalid. In the positive control cell pellet, at least 70% of the
cells containing cell membrane staining with at least 2+ intensity, and any background stain-
ing less than 1+ intensity are considered acceptable. In the negative cell pellet, the majority
of cells should demonstrate no staining, and any background staining should be less than
1+ intensity. The ideal positive inhouse NSCLC control tissue should provide the spectrum
of staining intensity from weak-to-moderate cell membrane staining, whereas the ideal
negative inhouse control should demonstrate no staining on the tumor cells except on the
tumor associated immune cells. All results for the patient specimen should be considered
invalid if the staining of control tissue is inappropriate. Formalin-fixed paraffin embedded
(FFPE) tonsil tissue can be used as an optional control with PD-L1 staining on the crypt
epithelium and follicular macrophages in the germinal centers, but not on the surface epi-
thelium. FFPE placental tissue is another control option, with PD-L1 staining observed in
syncitiotrophoblastic cells (Dolled-Filhart 2016).
The PD-L1 expression is evaluated by tumor proportion score (TPS), which is defined
as the percentage of viable tumor cells with at least partial membrane staining relative to
all viable tumor cells in the examined section (Garon 2015).
The evaluation of the scores includes partial or complete membrane staining (at least 1+
intensity) that is perceived distinct from cytoplasmic staining. Exclusive cytoplasmic stain-
ing should be excluded from the scoring; cytoplasmic staining is seen with membranous
staining in most instances. Only viable tumor cells are included in the scoring. All other
(stained) cells, such as tumor-associated immune cells, normal/non-neoplastic cells, and
necrotic cells, should be excluded from evaluation.
The scoring is interpreted as:
1. no PD-L1 expression (TPS<1%) (Figure 1);
2. PD-L1 expression (TPS 1-49%) (Figure 2); and,
3. high PD-L1 expression (TPS ≥ 50%) (Figure 3).
The tumor should be considered PD-L1 positive, and the patient eligible for KEYTRUDA®
(pembrolizumab) first-line therapy (Garon 2015) if the specimen shows high PD-L1 expres-
sion (TPS ≥ 50%), while at least PD-L1 expression (1-49% TPS) is required for treatment in
second-line or later.
The PD-L1 scoring is best evaluated on a representative tumor block from the surgical
resection specimen. Alternatively, staining can be undertaken on small biopsy specimens,
Figure 1. NSCLC stained with PD-L1 IHC 22C3 pharmDx

showing no expression (TPS < 1%) (40X magnification). Figure 2. NSCLC stained with PD-L1 IHC 22C3 pharmDx
showing low expression (TPS 1-49%) (40X magnification).
such as those obtained by bronchial or core
biopsies. Although staining on the cytology
specimens can be done, none of the trials
using clone 22C3 have validated these tests
to date, and the fixation procedures used in
the preparation of some cytology material
(eg, alcoholic fixation) may adversely affect
the performance of the assay. Suggested infor-
mation to include when reporting results with
PD-L1 IHC 22C3 pharmDx is provided in
Box 1.
Figure 3. NSCLC stained with PD-L1 IHC 22C3 pharmDx

Interpretation Pitfalls showing high expression (TPS 90%) (20X magnification).
A variety of pitfalls and artifacts (Figures
4-7), such as non-specific background, edge artifacts,
Box 1. Suggested Information to Include
crush artifacts, necrosis, or poor fixation, may be When Reporting Results from the PD-L1 IHC
encountered when evaluating the PD-L1 staining. 22C3 pharmDx Assay
As noted earlier, immune cells, including mac- General Information
rophages and lymphocytes, should be excluded • Positive control results (Pass/Fail)
• Negative control results (Pass/Fail)
from the scoring. Macrophages are usually present
• Adequate tumor cells (≥ 100 cells)
in intra-alveolar spaces or infiltrating within the are present (Yes/No)
tumor, and may show significant immunopositivity. • Tumor Proportion Score:
In addition, they may contain anthracotic or other PD-L1 Expression
pigments in the cytoplasm which may confound IHC • ___ None (< 1%)
interpretation. The small lymphoid cells with bare • ___ Low (1-49 %)
cytoplasm should be differentiated from the tumor • ___ High (≥ 50 %)
with their smaller and regular nuclei. Optional Information
• Presence/amount of tumor-associated
Predictive Significance (see also Chapter 2) immune cells
• PD-L1 positivity in increments of 10%
The value of the PD-L1 IHC 22C3 pharmDx assay in • Other comments to the clinician
predicting treatment response to pembrolizumab in
PD-L1 22C3 PHARMDX ASSAY 59
patients with NSCLC had been demonstrated in several large-scale clinical trials (Tables
2 and 3) (Baas 2016; Garon 2015; Herbst 2016; Hui 2016). In general, increased PD-L1
expression (higher TPS) is associated with a higher ORR and with favorable outcome. In
the initial KEYNOTE-001 phase 1 trial, which included both treatment-naïve and previ-
ously treated patients with NSCLC, the ORR was 10.7%, 16.5%, and 45.2% for patients
with a TPS of less than 1%, 1% to 49%, and 50% or greater, respectively (Garon 2015). The
Figure 4. NSCLC stained with PD-L1 primary antibody show- Figure 5. Pulmonary macrophages present in the alveo-
ing strong staining of the TAIC which should be excluded lar space with strong PD-L1 membrane staining should be
from the scoring (40X magnification). excluded from the scoring (40X magnification).
Figure 6. NSCLC stained with PD-L1 primary antibody show- Figure 7. NSCLC specimen stained with PD-L1 primary
ing moderate cytoplasmic staining of tumor cells, which antibody with the tumor cells showing a granular pattern.
should be excluded from the scoring (40X magnification). Only perceptible and convincing membrane staining can
be included in the scoring.
progression-free survival and overall survival were also better for patients with TPS of 50%
or greater, compared with those with TPS of <1% or 1% to 49% (Hui 2016). On the basis of
receiver-operating-characteristic (ROC) curves analysis, membranous PD-L1 expression in
at least 50% of tumor cells (TPS ≥ 50%) was selected as the cutoff in this study. Evaluation
of PD-L1 expression on immune cells did not further improve the predictive value of the
assay (Garon 2015).
The subsequent KEYNOTE-010 trial, a randomized phase 2/3 study, compared the effi-
cacy of pembrolizumab with standard of care treatment (docetaxel) for previously treated
advanced NSCLC testing positive for PD-L1 (defined as TPS ≥ 1%). Pembrolizumab was
superior to docetaxel in terms of overall survival and benefit-to-risk profile. In the subgroup
analysis stratified by extent of PD-L1 expression, gradual increases in the ORR and overall
survival were associated with higher TPS. The ORR was 8.6%, 15.8%, 22.6%, and 33.7% in
patients with a TPS of 1% to 24%, 25% to 49%, 50% to 74%, and ≥75%, respectively (Herbst
2016). Because this study included only PD-L1 positive (TPS ≥ 1%) NSCLCs, the efficacy of
Table 2. 22C3 Expression in Lung Cancer Samples
First Author Tumor Disease Tumor Proportion Score

Total
(year) Histology Stage <1% 1-24% 25-49% 50-74% 75-100%
Garon 2015 NSCLC IV 323(39.2%) 255(31.0%) 55(6.7%) 71(8.6%) 120(14.6%) 824
Herbst 2015 NSCLC IV 747 (33.6%) 842 (37.9%) 633 (28.5%) 2222
Cooper et al.
NSCLC I-III 487 (71.8%) 141 (20.8%) 50 (7.4%) 678
2015
Yeh et al. AdenoCA I-IV 182 (83.1%) 37 (16.9%) 219
2016
pembrolizumab versus docetaxel in PD-L1 negative (TPS < 1%) NSCLCs remained unde-
termined. In this study, pembrolizumab was superior to docetaxel regardless of whether
a recent or archival tumor sample was used for PD-L1 assessment, suggesting that either
contemporary biopsy samples or aged archival specimens are suitable for assessment.
In the first-line treatment setting, subgroup analysis of treatment–naïve patients in
the KEYNOTE-001 phase 1 trial also showed that the ORR and overall survival gradu-
ally increased with higher TPS. The ORR was 8.3%, 17.3%, and 51.9% for patients with
a TPS of less than 1%, 1% to 49%, and ≥ 50%, respectively (Hui 2017). The randomized
phase 3 KEYNOTE-024 trial also compared the efficacy of pembrolizumab with chemo-
therapy in previously untreated patients with advanced NSCLC whose tumors expressed
high levels of PD-L1 (defined as TPS ≥ 50%) and who had no sensitizing EGFR mutation or
ALK translocation. For this group of patients, pembrolizumab was associated with superior
progression-free and overall survival, with fewer adverse events compared with platinum-
based chemotherapy (Reck 2016). Another randomized phase 3 trial, the KEYNOTE-042
trial, is designed to evaluate the efficacy and safety of pembrolizumab compared with che-
motherapy as first-line therapy for PD-L1—positive advanced NSCLC (defined as TPS ≥ 1%).
PD-L1 expression (TPS ≥50% versus 1% to 49%) will be included among the randomization
stratification criteria (Mok 2016). The results are forthcoming.
There have also been ongoing clinical trials testing the efficacy and safety of pem-
brolizumab in combination with other therapies for advanced NSCLC. For example, the
KEYNOTE-021 trial evaluated the efficacy and safety of pembrolizumab plus chemotherapy
or ipilimumab, another immune checkpoint inhibitor targeting CTLA-4. Pembrolizumab
in combination with chemotherapy yielded substantial clinical efficacy, with an ORR of
55% compared with 28% for chemotherapy alone (Langer 2016). However, the combination
of pembrolizumab and ipilimumab was associated with significant toxicity, and the ORR
was similar to that of pembrolizumab alone (Gubens 2016). In contrast to pembrolizumab
Table 3. Summary of Treatment Results in NSCLC Clinical Trials Applying PD-L1 IHC 22C3 pharmDx Assay
PD-L1 No. of PFS (median), PD-L1 Predict
Trial Name Rx Line 2 Drug TPS Patients ORR, % OS (median), months months Trmt Resp? References
Pembrolizumab (Herbst
KEYNOTE 010 Treated 1-24% 471 8.6 (Pem) vs 10.9 (Doc) 9.7 (Pem) vs 8.5 (Doc) 2.6 (Pem) vs 4.0 (Doc) Yes
vs docetaxel (Doc) 2016)
25-49% 120 15.8 (Pem) vs 9.1 (Doc) 9.8 (Pem) vs 9.9 (Doc) 2.9 (Pem) vs 3.8 (Doc)
50-74% 158 22.6 (Pem) vs 9.6 (Doc) 15.8 (Pem) vs 8.2 (Doc) 4.3 (Pem) vs 4.3 (Doc)
≥75% 284 33.7 (Pem) vs 7 (Doc) 16.6 (Pem) vs 8.2 (Doc) 6.2 (Pem) vs 4.0 (Doc)
Treated and (Garon
KEYNOTE 001 Pembrolizumab <1% 28 10.7 Yes
Naïve 2015)
1-49% 103 16.5
≥50% 73 45.2
(Hui 2016;
KEYNOTE 001 Naïve Pembrolizumab <1% 12 8,3 14.7 3.5 Yes
Hui 2017)
1-49% 52 17.3 19.5 4.2
≥50% 27 51.9 Not reached 12.5
KEYNOTE 001 Treated Pembrolizumab <1% 90 9.9 8.6 Yes (Hui 2016)
1-49% 168 12.9 8.2
≥50% 138 38.3 15.4
Pembrolizumab 44.8 (Pem) vs 27.8 10.3 (Pem) vs 6.0
KEYNOTE 024 Naïve ≥50% 305 Not reached Not available (Reck 2016)
vs chemotherapy (chemotherapy) (chemotherapy)
Pembrolizumab + (Langer
KEYNOTE 021 Naïve <1% 21 57 No
chemotherapy 2016)
1-49% 19 26
≥50% 20 80
Pembrolizumab + (Gubens
KEYNOTE 021 Naïve <1% 21 19 17 6 No
ipilimumab 2016)
1-49% 18 33 Not reached Not reached
PD-L1 22C3 PHARMDX ASSAY
≥50% 6 17 2 1
Abbreviations: NSCLC, non–small cell lung cancer; TPS, Tumor Proportion Score; ORR, objective response rate; OS, overall survival; PFS, progression-free survival; Predict Trmt Resp, predictive of treatment response.
61
monotherapy, the treatment response of these combination regimens seemed to be indepen-

dent of PD-L1 expression. Further studies are warranted to confirm this finding (Langer 2016;
Gubens 2016).
Conclusion
The in vitro diagnostic PD-L1 IHC 22C3 pharmDx assay performed on the Dako Autostainer
Link 48 platform is an immunohistochemical assay for detection of PD-L1 protein expression
in advanced-stage NSCLC and for determination of eligibility for treatment with pembro-
lizumab, a humanized monoclonal IgG4 kappa isotype antibody against PD-1. The assay
assesses PD-L1 protein expression by evaluating TPS, which is the percentage of viable tumor
cells showing either partial or complete membrane staining. Increasing PD-L1 expression
(higher TPS) is generally associated with higher objective response rates and favorable
outcome in patients treated with pembrolizumab and the assay is approved as a companion
diagnostic assay for pembrolizumab.
Acknowledgement
The authors thank Dr. Yi-Chen Yeh, Dr. Shiou-Fu Lin, and Dr. Hsiang-Ling Ho for preparing
the text and micrographs of the chapter.
PD-L1 SP142 Assay
By Ross A. Soo, Bernadette Reyna Asuncion, and Reinhard Buettner 6
The Ventana PD-L1 (SP142) Assay (Ventana Medical Systems Inc.) is used to detect PD-L1
expression in tumor cells and immune cells in formalin-fixed, paraffin-embedded tissue.
The SP142 antibody clone has been used in clinical trials of patients with advanced-stage
non-small cell lung cancer (NSCLC), with scoring conducted using both tumor and immune
cells (Fehrenbacher 2016). It has been approved by the US Food and Drug Administration
as a complementary diagnostic tool to select patients with advanced urothelial carcinoma
or advanced NSCLC for atezolizumab therapy. The approval in urothelial cancer was based
on a phase II study in which programmed cell death ligand-1 (PD-L1) expression in 5%
or greater of immune cells was associated with increased objective response for patients
treated with atezolizumab (Rosenberg 2016). Regarding NSCLC, PD-L1 expression in at
least 50% of viable tumor cells or in at least 10% of viable immune cells has been associated
with enhanced overall survival with atezolizumab based on two trials, the phase III OAK
trial (Rittmeyer 2017) and the phase II POPLAR trial (Fehrenbacher 2016).
The Ventana PD-L1 (SP142) Assay is a rabbit monoclonal anti–PD-L1 antibody, that
recognizes the intracellular domain of the PD-L1 protein ligand. In the United States, the
Ventana PD-L1 SP142 assay is approved only for use on the BenchMark ULTRA (Ventana
Medical Systems Inc.) platform, with the OptiView DAB IHC Detection Kit (Ventana Medical
Systems Inc.) and the OptiView Amplification Kit (Ventana Medical Systems Inc.). Outside
the US, the assay is approved for use on the Ventana Ultra, GX and XT platforms’.
Three serial sections for testing are required from each case: the first for hematoxylin
and eosin staining, a second for negative reagent control staining, and a third section for
SP142 assay staining. The recommended control for use with this assay is tonsil tissue, which
should be used as both a positive and a negative control for each staining run to monitor
the performance of processed samples, as well as to test reagents and instruments. Control
tissue should be fixed as soon as possible and processed in the same way as patient tumor
samples. Tonsil tissue contains positive and negative staining epithelial and immune cells,
which are used to confirm if the assay performed appropriately. A matched negative reagent
control slide using the Rabbit Monoclonal Negative Control Ig (Ventana Medical Systems
Inc.) antibody should be conducted for each run to assess for nonspecific staining. Use of a
different negative control reagent may cause false results.
Evaluation of Staining
Tonsillar Tissue Controls
Acceptable tonsil staining should show moderate-to-strong PD-L1 staining in lymphocytes
and macrophages in the germinal centers, whereas reticulated crypt epithelial cells should
show diffuse staining. Generally, there should not be any PD-L1 expression in immune cells
in the interfollicular regions and on superficial squamous epithelium, however, rare PD-L1
positive immune cells may be found. Unacceptable staining in controls includes excessive
nonspecific background staining that would conceal PD-L1–positive cells. Unacceptable
staining may include weak-to-none PD-L1 staining in lymphocytes and macrophages in
germinal centers, as well as in reticulated crypt epithelial cells. If the tissue control does not
display the appropriate staining, patient samples should not be considered for evaluation,
and staining should be repeated.
Caution should be exercised when interpreting the staining intensity of tumor cells
because of the strong PD-L1 staining in control tissues, and controls should not be used as
an aid in formulating a specific PD-L1 expression score. Further data is needed regarding
use of samples that reflect the tumor entity, such as with urothelial cancer or NSCLC.
If nonspecific staining is present, it often has a diffuse appearance that may be identified
using the negative reagent control slide stained with Rabbit Monoclonal Negative Control
Ig. Intact cells should be used for interpretation of staining results. If background staining
is excessive, the interpretation of the patient samples should be considered invalid.
Staining Patterns for Tumor and Immune Cells

The sample should contain at least 50 tumor cells with associated stroma for tumor-cell
staining results to be valid; necrotic areas should be avoided. SP142 differs significantly
from other PD-L1 assays by its distinct staining pattern, revealing both membranous and
granular cytoplasmic staining in tumor cells (Scheel 2016). This contrasts with all other
antibody clones, which show cell surface-defined linear membranous staining. The distinct
pattern for the SP142 assay relies on detection by the Ventana OptiView system and the
Ventana BenchMark Ultra platform. Linear membranous staining is shown when the anti-
body from the SP142 kit is used in combination with other detection systems (Scheel et al,
submitted).
PD-L1–positive tumor cells usually show partial or fully circumferential membranous
staining. When these tumor cells are stained with the Ventana OptiView system, the cells
may show cytoplasmic granular staining of variable intensity. However, only membranous
staining is considered for scoring. This cytoplasmic staining is independent from mem-
branous staining, and both types of positive tumor cells should be evaluated. When using
the brown 3,3'-diaminobenzidine color for visualization, it is important not to confuse
immunohistochemistry (IHC) signals with anthracotic or iron pigment. Another pitfall is
the intense PD-L1 staining on normal bronchiolar epithelium over lymphoid tissue. This
bronchus-associated lymphoid tissue-like staining must not be confused with specific tumor-
cell staining.
PD-L1 SP142 ASSAY 65
A population of immune cells, such as lymphocytes, macrophages, dendritic cells, and

neutrophils, stain for PD-L1. The immune cells are typically found in the intratumoral and
peritumoral (invasive margin) regions (Figure 1). In addition, focal or diffuse scattered single
immune cells or small aggregates of cells found in the intratumoral stroma, peritumoral
stroma, or both might be observed. Circumferential immune-cell membrane staining also
is observed. Immune-cell staining can be seen as fine-punctate or diffuse-granular staining
in neutrophils.
Discrimination between tumor and immune cells can be challenging and relies on conven-
tional characteristics of tumor cells, such as enlarged or atypical nuclei and clear epithelial
differentiation (e.g., formation of acini). It might not be possible to distinguish scattered
single tumor cells between intense patches of immunopositive immune cells, especially
in cases with high immune-cell expression scores. Similarly, positive immune cells can be
impossible to identify if admixed with strongly positive tumor cells. A careful comparison
of immunostained slides with the corresponding slides of the same tumor area that were
stained with hematoxylin and eosin is necessary to help identify immune cells mixed among
tumor cells. In addition, a high-magnification review of the PD-L1–stained slide may also
assist in differentiating between tumor-cell and immune-cell staining.
A1 A2 B
Intravascular
A3 A4
Intratumoral
C D1 D2
PD-L1+
lymphocytes
(brown)
Anthracotic
pigments
(black)
Figure 1. Staining with the Ventana PD-L1 (SP142) Assay demonstrating: (A) tumor-cell membranous staining, (B)
intravascular programmed cell death ligand-1 (PD-L1)– positive immune cells and intratumoral lymphocytes at the
squamocolumnar junction, (C) lymphoid aggregates with co-existing anthracotic pigments and PD-L1– positive
immune cells, and (D) tumor cells (encircled) with partial-membrane staining. The left side of the bottom right image
(D1) shows darkly stained necrotic cells, which may be tumor cells and/or immune cells. Darkly stained cells are dif-
ficult to assess. The right side of the same image (D2) shows partial tumor-membrane staining. 40x magnification.
Scoring, Reporting, and Interpretation

Sections are scored using a stepwise approach based on the criteria outlined in Figure 2.
First, the stained slides are assessed for tumor-cell staining. If the specimen contains
any discernible PD-L1 membrane staining of any intensity in at least 50% of tumor cells,
Step 1: Assessment of PD-L1 staining in

tumor cell (TC) TC ≥ 50% Report as TC ≥ 50%
What is the PD-L1 expression in TC membrane?
TC < 50%
Step 2: Assessment of PD-L1 staining in IC ≥ 10% Report as TC < 50%, IC ≥ 10%

tumor infiltrating immune cells (IC)
What proportion of the tumor area* is occupied
by PD-L1 positive IC of any intensity? IC < 10% Report as TC < 50%, IC < 10%
* Tumor area is defined as tumor cells, associated intratumoral

and contiguous peritumoral stroma
Figure 2. Stepwise scoring algorithm for programmed cell death ligand-1 (PD-L1) expression in non-small cell
lung cancer samples using the Ventana PD-L1 (SP142) Assay (approach approved by US FDA). *Tumor area is
defined as tumor cells (TC), associated intratumoral and contiguous peritumoral stroma. IC = immune cells.
the sample is assigned a PD-L1 expression level of 50% or greater. If the specimen shows
staining in less than 50% of tumor cells present, immune-cell staining is then assessed. If
the sample contains PD-L1 staining of any intensity in immune cells occupying at least
10% of the tumor area, the case will be given a PD-L1 expression level of greater than 10%
for immune cells. If the specimen contains PD-L1 staining of any intensity in immune cells
covering less than 10% of the tumor area, the case will be given a PD-L1 expression level
of less than 10% for immune cells.
In the clinical trials, the tumor-cell scoring consisted of TC0 (defined as less than 1% of
tumor cells expressing PD-L1), TC1 (1% to <5%) TC2 (5% to <50%) and TC3 (50% or more).
In addition, immune cells were scored as IC0 to IC3, where IC0 is defined as<1% of PD-L1
tumor immune cells, IC1 (1 to less than 5%), IC2 (5-less than 10%) and IC3 (10% or more),
depending on the percent of immune cells expressing PD-L1 (Table 1).
Tumor cells are scored as the proportion of viable tumor cells showing PD-L1 membrane
staining of any intensity (Figure 1A). Tumor necrosis is excluded from scoring. Stroma that
is part of tissue fragment from small biopsies (in which samples often might be fragmented)
but not contiguous to viable tumor is excluded. Only stroma that is contiguous to individual
tumor nests is included in the tumor-area definition. Positive staining includes partial or
circumferential membrane staining (Figure 1D2) and weak or intense membranous staining.
The immune cells are scored using the proportion of the tumor area that is occupied by
PD-L1–positive immune cells of any intensity (Figure 1B). The tumor area is defined as the
area occupied by viable tumor cells and by their associated intratumoral and contiguous
peritumoral stroma. Necrotic tumor is excluded from this definition of tumor area. In frag-
mented tissue samples, such as from biopsies, only stroma that is contiguous to individual
tumor nests is included in the definition of tumor area; stroma that is part of a tissue frag-
ment but not contiguous to viable tumor is excluded. Of note, any PD-L1 staining, regardless
of the type of immune cell or its location, is included, excluding alveolar macrophages. The
typical procedure of immune-cell scoring is summarized in Figure 3.
Table 1. Reported Distribution of Lung Cancer Samples Using the SP-142 Assay and Different Programmed Cell Death Ligand-1 (PD-L1) Expression Cut-offs
First Author Tumor Disease Tumor Cell (TC) Expression Immune Cell (IC) Expression TC1/2/3 or TC2/3 or TC3 or Total*
(Year) Histology Stage Cut-off (Score) Cut-off (Score) IC1/2/3 IC2/3 IC3
>1% and ≥5% and ≥1% and ≥5% and

<5% (TC1) <50% (TC2) ≥ 50% (TC3) <5% (IC1) <10% (IC2) ≥ 10% (IC3)
Fehrenbacher NSCLC Advanced, post– 109 (38.0%) 69 (24.0%) 30 (10.5%) 162 (56.4%) 55 (19.1%) 18 (6.3%) 195 (68%) 105 47 287
(2016) platinum-based (37%) (16%)
chemo-therapy
Ritmeyer NSCLC Advanced, post– NA NA NA NA NA NA 463 (54%) 265 137 842
(2017) platinum-based (31%) (16%)
chemo-therapy
Besse NSCLC Advanced NA NA NA NA NA NA NA 659 302 659

(2015) (100%) (46%)
Spigel NSCLC Advanced NA NA NA NA NA NA NA 114 41 114

(2015) (100%) (36%)
Ilie NSCLC Resected 23% NA NA 79% NA NA 118 (74%) NA 38% 160

(2016) surgical sample,
IB-IIIB
*Number of cases evaluable for PD-L1 expression. NSCLC = non-small cell lung cancer, NA = not available
PD-L1 SP142 ASSAY
67
Figure 3. Scoring for programmed cell death ligand-1 (PD-L1) tumor-cell aggregate staining. (Left) The tumor percentage
area (blue) should be determined using a minimum of 50 viable tumor cells. (Middle) The immune cell percentage area
(encircled red) within the tumor. (Right) Estimate the proportion of tumor occupied by immune-cell aggregates. The H&E
stained section may assist in assessing the tumor area.
As previously discussed, it is frequently difficult to determine tumor areas in samples

from biopsies because the tissues are fragmented. The stroma adjacent to the tumor nests,
where positive PD-L1 immune cells are scored, should be considered as the tumor area when
estimating the expression score of the whole tumor (Figure 4).
Immune cells (blue boundary)
Tumor area (red)
Tumor area (red boundary)
Figure 4. Tumor biopsies with less than 50% of tumor cells and greater than 10% of immune cells showing
programmed cell death ligand-1 (PD-L1) expression.
Suggested information to include when reporting results with Ventana PD-L1 (SP142)
Assay is found in Box 1.
Interpretation Pitfalls
As with all the assays available for PD-L1 expression testing, a variety of pitfalls and artifacts
(e.g., nonspecific background, edge artifacts, crush artifacts, necrosis, or poor fixation) might
be encountered when evaluating PD-L1 staining with the Ventana PD-L1 (SP142) Assay (see
Chapter 3 for details). Some example images of staining artifacts specific to SP142 from chal-
lenging cases are shown in Figure 1. Of note, intravascular immune cells, PD-L1–positive
immune cells within blood vessels in the tumor stroma, are not considered for immune-cell
scoring (Figure 1B). In addition,

Box 1. Suggested Information for Inclusion When Reporting Results
anthracotic deposits should not be from the VENTANA PD-L1 (SP142) Assay
confused with PD-L1 positivity or
General Information
subsequently scored (Figure 1C).
Lymphoid aggregates with coexist- • Negative control results (Pass/Fail)
ing anthracotic pigments can also • Whether adequate tumor cells (≥50 cells) are present (Yes/No)
be challenging (Figure 1C), and PD-L1 IHC SP-142 Result to Clinician*
darkly stained cells (Figure 1D1) • Tumor cell expression ≥ 50% ___
are difficult to assess. • Immune cell expression ≥10% (and tumor cell
expression < 50%) ___
• Tumor cell expression < 50%, Immune cell expression
SP-142 Expression in Lung
< 10% ___
Cancer Samples
• Expression status
PD-L1 expression detection using
° Tumor cell % (___< 1%, ___≥ 1%, ___≥ 5%, ___≥ 50%)
SP142 has been reported in several
° Immune cell % (___< 1%, ___> 1%, ___> 5%, ___> 10%)
studies (Table 1). The frequency of
cases with >=50% expression for Additional Information
• Other comments
tumor cells (TC3) or >= 10% tumour
area infiltrated by immune cells (IC3 in the randomized trials OAK (Rittmeyer 2017) and
POPLAR (Fehrenbacher 2016) was approximately 16%. The frequency of this level of expres-
sion was higher in the BIRCH (Besse 2015) and FIR (Spigel 2015) trials due to selection bias;
only patients with PD-L1 expression of 5% or greater (TC2/3) in tumor and/or immune cells
(IC2/3) were eligible for treatment.
Predictive Significance
Several studies have reported the association between PD-L1 expression in tumor and
immune cells using clinical outcomes for patients with advanced-stage NSCLC who were
treated with atezolizumab. Atezolizumab is active in a range of solid tumors, as shown in
a phase I study (Herbst 2014). The objective overall response rate was 23% for the entire
NSCLC cohort; however, this increased to 83% when PD-L1 expression positivity was
scored as TC3 or IC3 (Table 2). In addition, there was an association between responses and
PD-L1 expression in tumor-infiltrating immune cells for patients with NSCLC (p = 0.015)
and with all tumor types (p = 0.007), but there was no association between response and
tumor-cell PD-L1 expression for patients with NSCLC (p = 0.920) and with all tumor types
(p = 0.079).
Overall survival was longer for patients treated with atezolizumab (HR: 0.73 95% CI:
0.53–0.99; p = 0.04) in the POPLAR study, a phase II study of patients who had prior treat-
ment with a platinum-based chemotherapy doublet and who were randomly assigned to
treatment with atezolizumab or docetaxel (Fehrenbacher 2016). Of note, increased PD-L1
expression on tumor cells and tumor-infiltrating immune cells was independently predic-
tive of improved overall survival with atezolizumab. The overall survival was similar in
the atezolizumab and docetaxel groups (HR: 1.04; 95% CI: 0.62-1.75; p = 0.871) for patients
with TC0 and IC0 PD-L1 expression in immune cells. Furthermore, IC PD-L1 expression
was associated with T-effector and interferon-γ gene signature, suggesting pre-existing
immunity in the tumor.
Table 2. Overall Survival, Progression-free Survival, and Objective Response Rate According to Tumor- and
Immune-Cell Subgroups in Patients Treated with Atezolizumab With or Without Docetaxel
Phase I
ClinicalTrials.
Phase III Phase II Phase II gov: Phase II
OAK Trial POPLAR Trial FIR Trial NCT01375842 BIRCH Trial
PD-L1
(Rittmeyer 2017) (Fehrenbacher 2016) (Spigel 2015) (Herbst 2014) (Besse 2015)
Expression
Score* Atezolizumab Docetaxel Atezolizumab Docetaxel Atezolizumab Atezolizumab Atezolizumab
Objective Response Rates (%)
Overall 14 13 14.6 14.7 NA Overall IC: 23 NA
TC3 or IC3 31 11 37.5 13.0 First-line: 29 IC3: 83 First-line: 26
Second-line+, Second-line: 24
no CNS mets: 24
Third-line+: 27
Second-line+,
treated CNS
mets: 25
TC2/3 or NA NA 22.0 14.5 First-line: 26 IC2: 14 First-line: 19
IC2/3
Second-line+, Second-line: 17
no CNS mets: 16
Third-line+: 17
Second-line+,
treated CNS
mets: 23
TC1/2/3 or 18 16 18.3 16.7 NA IC1: 15 NA
IC1/2/3
TC0 and IC0 8 11 7.8 9.8 NA IC0: 20 NA
Progression-free Survival (PFS; [HR; 95% CI]) 6-month PFS
Overall 2.8 months 4.0 months 2.7 months 3.0 months NA IC overall: NA
(HR: 0.95; (HR: 0.94; 15 weeks
0.82-1.10; 0.72-1.23:
p =0.4928) p = 0.645
TC3 or IC3 4.2 months 3.3 months 7.8 months 3.9 months First-line: 5.4 IC3: NE First-line: 48%
(HR = 0.63) (HR: 0.60; months
0.31-1.16: Second-line:
p = 0.127 Second- line+, 34%
no CNS mets: 4.1
months Third-line+:
39%
Second-line+,
treated CNS
mets: 2.3 months
TC2/3 or NA NA 3.4 months 2.8 months First-line: 4.5 IC2: 11 weeks First-line: 46%
IC2/3 (HR: 0.72; months
0.47-1.10: Second-line:
p = 0.124) Second-line+, 29%
no CNS mets:
2.7 months Third-line+:
31%
Second-line+,
treated CNS
mets: 2.5 months

TC 1/2/3 or 2.8 4.1 months 2.8 months 3.0 months NA IC1: 6 weeks NA
IC 1/2/3 (HR = 0.91) (HR: 0.85;
0.63-1.16;
p = 0.309)
TC0 and IC0 2.6 4.0 months 1.7 months 4.1 months NA IC0: 13 weeks NA
(HR = 1.0) (HR: 1.12;
0.72-1.77;
p = 0.611)
Overall Survival (OS; [HR; 95% CI]) 6-month OS
Overall 13.8 9.6 months 12.6 months 9.7 months NA NA NA
(HR: 0.73; (HR: 0.73;
0.62-0.87: 0.53-0.99:
p = 0.0003) p = 0.04
TC3 or IC3 20.5 8.9 months 15.5 months 11.1 First-line: NR NA First-line: 79%
(HR: 0.41; months
0.27-0.64: (HR: 0.49; Second-line+, Second-line:
p = 0.0001) 0.22-1.07: no CNS mets: NR 80%
p = 0.068)
Second-line+, Third-line+:
treated CNS 75%
mets: 7 months
TC2/3 or NA NA 15.1 months 7.4 months First-line: NR NA First-line: 82%
IC2/3 (HR: 0.54;
0.33-0.89: Second-line+, Second-line:
p = 0.014) no CNS mets: 76%
10.6 months
Third-line+:
Second-line+, 71%
treated CNS
mets: 6.8 months
TC1/2/3 or 15.7 months 10.3 months 15.5 months 9.2 months NA NA NA
IC1/2/3 (HR: 0.74; (HR: 0.59;
0.58-0.93: 0.40-0.85:
p = 0.0102) p = 0.005)
TC0 and IC0 12.6 months 8.9 months 9.7 months 9.7 months NA NA NA
(HR: 0.75; (HR: 1.04;
0.59-0.96: 0.62-1.75:
p = 0.0205) p = 0.871)
* Tumor-cell (TC) and immune-cell (IC) scoring = TC0, less than 1% of tumor cells expressing programmed cell death ligand-1 (PD-L1);
TC1, 1% to 5%; TC2, 5% to 50%; and TC3, greater than 50%. Abbreviations: NA, not applicable; CNS, central nervous system; mets,
metastasis; HR, hazard ratio; CI, confidence interval; NE, not estimable; NR, not reported.
The objective response rate ranged from 17% to 27% in the phase II BIRCH study (Besse
2015). In this study, patients had advanced-stage NSCLC and PD-L1 tumor or immune cell
expression of 5% or greater. Patients also had received prior treatment with atezolizumab
as first-line or subsequent therapy. Greater PD-L1 expression was associated with improved
responses. Similarly, the objective response rate was 16% to 26% in a phase II study (the FIR
trial) of patients with advanced NSCLC, enriched for PD-L1 expression in both tumor and
immune cells and with or without treated brain metastasis. In addition, patients had been
pre-treated or treated with atezolizumab in the first-line setting (Spigel 2015). In pre-treated
patients, increased PD-L1 expression (50% or greater) was associated with an increased
objective response rate and longer progression-free survival time, as well as increased land-
mark progression-free and overall survival rates.
The phase III OAK trial reported overall survival in the overall population of 13.8 months
for atezolizumab compared with 9.6 months for docetaxel (HR: 0.73 95% CI: 0.62-0.87; p =
0.0003); all patients had advanced pre-treated NSCLC and were randomly assigned to either
therapy (Rittmeyer 2017). Furthermore, atezolizumab conferred a survival benefit regardless
of PD-L1 expression status, with extremely similar hazard ratios for PD-L1–positive (HR =
0.74) and negative (HR = 0.75) patients. Figure 5 illustrates a case of a heavily pre-treated
patient with at least TC2/IC2 PD-L1 expression whose disease responded to atezolizumab.
A1 A2
B1 B2
Figure 5. Images of (A) left supraclavicular lymph nodes and (B) coeliac axis lymph nodes at
(1) baseline and (2) after two cycles of atezolizumab.
Conclusion
The Ventana PD-L1 (SP142) Assay is used to detect PD-L1 expression in both tumor and
immune cells as predictive marker for azetolizumab therapy. The sample should have at
least 50 tumor cells with associated stroma, and PD-L1 is expressed as membranous and
granular cytoplasmic staining in these cells. The SP142 assay is performed using a step-
wise approach. Tumor cells are scored by determining the percentage of area covered by
PD-L1–positive viable tumor cells and associated intratumoral and contiguous peritumoral
stroma. Immune cells are scored by determining the proportion of the tumor area that is
occupied by PD-L1–positive immune cells of any intensity. An association between clini-
cal outcomes and PD-L1 expression in the tumor and immune cells have been reported in
studies of patients with advanced-stage NSCLC treated with atezolizumab.
PD-L1 SP263 Assay
By Sanja Dacic and Arne Warth 7
The VENTANA PD-L1 (SP263) Rabbit Monoclonal Primary Antibody assay was developed
by Ventana Medical Systems, Inc., a subsidiary of Roche, in collaboration with AstraZeneca
for use with the VENTANA Benchmark ULTRA staining platform (Ventana Medical Systems,
Inc.). This assay detects PD-L1 expression and helps determine patient eligibility for treat-
ment with the immunotherapeutic drug durvalumab (Rebelatto 2016). Durvalumab is a
human monoclonal antibody directed against programmed cell death ligand-1 (PD-L1).
PD-L1 expression enables tumors to evade detection by the immune system through binding
to the programmed cell death-1 protein (PD-1) on cytotoxic T lymphocytes (Stewart 2015).
Durvalumab blocks PD-L1 interaction with both PD-1 and CD80 on T cells, countering the
tumor’s immune-evading tactics. Non-small cell lung cancer (NSCLC) cells with PD-L1
expression of at least 25% were analyzed using this assay in two recent clinical trials (Stewart
2015, Garon 2015). In a multicenter phase Ib study, however, durvalumab demonstrated anti-
tumor activity regardless of associated PD-L1 expression status (Antonia 2016). Recently,
the SP263 assay has been commercialized for identification of patients with non-squamous
cell NSCLC who are most likely to benefit from nivolumab (Chapter 4 discusses another
complementary diagnostic tool for nivolumab therapy, the PD-L1 IHC 28-8 pharmDx Assay.)
VENTANA PD-L1 (SP263) Rabbit Monoclonal Primary Antibody is a rabbit monoclonal
primary antibody produced against PD-L1 that localizes to and stains the membranous
and/or cytoplasmic regions of cells. Anti–PD-L1 SP263 binds to an epitope corresponding
to amino acids 284-290 of the PD-L1 protein (Quon 2016). The SP263 assay is intended for
laboratory use for the detection of the PD-L1 protein in formalin-fixed, paraffin-embedded
tissue, and the marketed package includes 50 tests. Acceptable fixatives also include zinc
formalin and Z-5 fixatives. Fixatives not recommended for use are: 95% alcohol; alcohol,
formalin, and acetic acid (AFA); and Prefer (Anatech LTD). As previously mentioned, the
assay was optimized for the automated VENTANA Benchmark ULTRA platform. Detection
is optimized with the OptiView DAB IHC Detection Kit (Ventana Medical Systems, Inc.),
which is an indirect, biotin-free system for detecting mouse immunoglobulins G and M, as
well as rabbit primary antibodies. The slides should be stained immediately because the
antigenicity of cut tissue sections might diminish over time. Placenta is recommended as a
positive control because the SP263 assay demonstrates a uniform staining of the membrane
and/or cytoplasm in trophoblast lineage cells that is moderate (2+)-to strong (3+).

Durvalumab
The assay should be evaluated and scored by a qualified pathologist using light microscopy.
The interpretation of PD-L1 staining must be complemented by the evaluation of slides
stained with hematoxylin and eosin and of negative control reagent-stained slides. SP263
staining is interpreted as positive if membranous and/or cytoplasmic protein expression at
any intensity greater than background staining is detected in at least 25% of tumor cells. At
least 100 viable tumor cells should be scored. Data regarding the assessment of the staining
intensity has not been presented. Tumor-infiltrating lymphocytes and intra-alveolar mac-
rophages can also demonstrate linear membranous, diffuse cytoplasmic, and/or punctate
staining. Staining of the tumor-infiltrating lymphocytes, non-neoplastic cells, and necrotic
cells is not included in the scoring criteria.
The staining may appear heterogenous in both small biopsy and large resection speci-
mens. The staining can be very strong and homogenous (Figure 1A); however, some tumors
show heterogenous expression. Variable-intensity strong and moderate staining is easily
identified at low power magnification (2x or 4x), whereas weak staining is best seen at
high-power magnification (20x). The “histo” (H) score, which considers both the percent-
age of tumor cells that stain positively and staining intensity, has not been used in the
clinical trials. Heterogeneity of staining might be, in part, caused by pre-analytical factors.
Therefore, staining of freshly cut tissue on new charged (“plus”) slides is recommended
to limit exposure to room humidity and to minimize plastic exposure (i.e., use of glass or
metal containers and racks).
Nivolumab
Recently, the SP263 assay has become available in Europe for the identification of patients
with non-squamous cell NSCLC who are most likely to benefit from nivolumab. This was
based on high concordance between PD-L1 expression status as determined by the PD-L1
IHC 28-8 pharmDx (Agilent Technologies/Dako) and VENTANA PD-L1 (SP263) Rabbit
Monoclonal Primary Antibody assays, although use of the SP263 assay tends to result in
more intense staining for carcinoma and immune cells than the 28-8 and PD-L1 IHC 22C3
pharmDx (Agilent Technologies/Dako) assays (Scheel 2016, Hirsch 2017). When evaluat-
ing for nivolumab therapy, PD-L1 expression status should be scored using membranous
staining of tumor cells. Scoring subgroups include: less than 1%, 1% to 5%, 5% to 10%, and
10% or greater.
Reporting
Suggested information to include when reporting results with the SP263 assay, for both
durvalumab and nivolumab therapy consideration, is shown in Box 1.
Box 1. Suggested Information for Inclusion When Interpretation Pitfalls

Reporting Results from the VENTANA PD-L1 (SP263) A variety of pitfalls and artifacts including
Rabbit Monoclonal Primary Antibody Assay
nonspecific background, edge artifacts,
General Information crush artifacts, necrosis, or poor fixation
might be encountered when evaluating
• Negative control results (Pass/Fail)
• Whether adequate tumor cells (≥ 100 cells) the PD-L1 staining (see Chapter 3). There
are present (Yes/No) is concern that sampling error may result
• Tumor Proportion Score in discrepant results between different
PD-L1 IHC SP263 Result to Clinician specimen types (i.e., biopsy vs. resection),
• For durvalumab and studies are needed to address this
< 25% Expression issue. It has been shown for other PD-L1
≥ 25% Expression clones that PD-L1 expression is affected by
• For nivolumab
radiation and chemotherapy, which may
Expression < 1%
have been administered after a biopsy was
Expression ≥ 1% to < 5%
obtained, but similar data is not available
Expression ≥ 5% to < 10%
for SP263 (Sheng 2016).
Expression ≥ 10%
Many artifacts may lead to false-
Additional Information positive staining if staining is not carefully
• Information about tumor-associated immune cells
evaluated. Alveolar macrophages—par-
• Other comments to the clinician ticularly smoker’s macrophages—show
cytoplasmic granular staining (Figure
1B). Non-neoplastic lung parenchyma can show staining of inflammatory and interstitial
stromal cells. Extracellular mucin and tumor necrosis can show granular staining (Figure
1C). The staining of papillary adenocarcinoma should be interpreted with caution because
stromal cells in the fibrovascular cores may show moderate-to-strong staining that can be
interpreted as false-positive tumor-cell staining (Figure 1D). Red blood cells can show a
diffuse moderate staining.
Predictive, Prognostic Significance

In a phase I study by Rizvi et al. the response rate to durvalumab was 27% for patients with
PD-L1–positive NSCLC but only 5% for patients who were PD-L1 negative (Table 1; Rizvi
2015). In a phase II study of durvalumab in pre-treated patients with NSCLC, overall response
rates were 7.5% for PD-L1 expression of less than 25%, 16.4% for PD-L1 expression of 25%
to 90%, and 30.9% for PD-L1 expression of 90% or greater. Progression-free survival for the
same three groupings was 1.9, 3.3, and 2.4 months, respectively. One-year survival rates
were 34.5%, 47.7%, and 50.8%, respectively (Garassino 2017). Response was noted in 22%
of PD-L1–positive and 29% of PD-L1–negative tumors in a multicenter, nonrandomized,
open-label, phase Ib study of durvalumab plus tremelimumab in NSCLC (Table 1; Antonia
2016). Durvalumab in combination with gefitinib showed 77.8% to 80% objective response
rate in patients with NSCLC and with sensitizing EGFR mutations who had not received
prior therapy with a tyrosine kinase inhibitor. However, 55% of patients had grades 3 to 4
treatment-related toxicities (Gibbons 2016).
The prognostic significance of PD-L1 expression as detected by the SP263 assay is uncer-
tain. A recent univariate analysis demonstrated a correlation between improved overall
A B
C D
Figure 1. PD-L1 staining. A) An adenocarcinoma sample showing a diffuse, strong, and uniform staining with the VENTANA
PD-L1 (SP263) Rabbit Monoclonal Primary Antibody assay (20X magnification). B) The adenocarcinoma sample is negative
for programmed cell death ligand-1 (PD-L1) expression. Background staining shows normal lung parenchyma and includes
alveolar septa and smoker’s macropahges within airspaces. C) A lymph node specimen showing negative viable tumor cells.
Occasionally viable lymphocytes are positive for programmed cell death ligand-1 (PD-L1) expression. Necrotic debris shows
weak-to-strong focal staining (20X). D) A papillary adenocarcinoma sample showing strong staining of stromal cells and
inflammatory cells within fibrovascular cores. Tumor cells are mostly negative for programmed cell death ligand-1 (PD-L1)
expression (20X magnification).
Table 1. Summary of Clinical Trials with the Response Data

Overall Response
Rate/Positive (%)
Median Median
Study Treatment PD-L1 PD-L1 PFS OS
Drugs Design Line Positive Negative (months) (months) Clinical Trial Reference
Durvalumab Phase I ≥2 23/84 5/92 NA NA NCT01693562 Rizvi 2015
(AstraZeneca/ (27%) (5%)
MedImmune LLC)
Durvalumab plus ≥2
Tremelimumab Nonran- 2/9 4/14 (29%) NA NA NCT02000947 Antonia
(AstraZeneca/ domized, (22%) 2016
MedImmune LLC) phase Ib
Durvalumab plus
Gefitinib Phase I ≥2 NA NA NA NA NCT02088112 Gibbons
(AstraZeneca/ 2016
MedImmune LLC)
Durvalumab Phase 2 ≥3 16.4%* 7.5%** NA 10.9 NCT02087423 Garassino
(AstraZeneca) 2017
PD-L1 = programmed cell death ligand-1, PFS = progression-free survival, OS = overall survival, NA = not applicable.
* PD-L1 high (positive) was defied as ≥25% of tumor cells with membrane staining; cohort with PD-L1 expression ≥90% showed
overall response rate of 30.9%
** PD-L1 low/negative was defined as <25% of tumor cells with membrane staining
survival and SP263 results in basaloid squamous cell carcinoma. However, this correlation
was not confirmed in the multivariate analysis (Ilei 2016).
Conclusion
The VENTANA PD-L1 (SP263) Rabbit Monoclonal Primary Antibody assay is intended to
be used on the Ventana BenchMark ULTRA immunohistochemical stainer for detection
of PD-L1 expression in patients with NSCLC and other tumor types for treatment with
durvalumab. PD-L1 tumor cell expression of at least 25% was the standard requirement
in associated clinical trials. Recently, a Ventana SP263 assay has been commercialized for
identification of patients with non-squamous cell NSCLC who are most likely to benefit
from treatment with nivolumab.
PD-L1 73-10 Assay
By Mari Mino-Kenudson, Arne Warth, and Yasushi Yatabe 8
The immunohistochemistry (IHC) method is being developed by Dako (Agilent Technologies),
to detect programmed cell death ligand-1 (PD-L1) expression as a clinical decision-making
tool regarding use of the immunotherapeutic drug avelumab. The Dako PD-L1 IHC 73-10
Assay (previously known as PD-L1 IHC MSB0010718C assay) includes a primary recombi-
nant rabbit monoclonal antibody clone 73-10 that is a proprietary antibody of Merck KGaA
and is used by Dako under license. Avelumab is a fully human anti-PD-L1 immunoglobin G1
monoclonal antibody. By inhibiting PD-L1 interactions, avelumab is thought to enable activa-
tion of T cells and the adaptive immune system. By retaining a native fragment crystallizable
(Fc) region, avelumab also is thought to engage the innate immune system and may induce
antibody-dependent cell-mediated cytotoxicity. In clinical trials, patients with non-small cell
lung cancer (NSCLC) exhibiting PD-L1 expression of at least 1% of tumor cells as confirmed
by this platform appear to have improved progression-free survival and/or overall survival
(Gulley 2015, Verschraegen 2016).
Antibody Characteristics and IHC Procedure

The Dako PD-L1 IHC 73-10 assay uses the antibody clone 73-10 that was produced by
immunizing rabbits with a C-terminal cytoplasmic peptide of PD-L1. The test kits were
optimized to detect PD-L1 expression in formalin-fixed, paraffin-embedded human tissue
biopsy and surgical resection specimens as follows. The effects of different fixatives, such
as alcohol, on specificity and sensitivity of staining is unknown. A histologic section on
a positively charged slide prepared from the tumor specimen is first incubated with the
primary antibody clone 73-10 and then with a horseradish peroxidase (HRP) visualization
reagent. Subsequently the chromogen diaminobenzidine is added, which is oxidized by HRP
to a visible reaction product at the antigen site. Expression of PD-L1 is detected as brown
deposits on the membranes of and/or within the cytoplasm of positive cells.
According to a Merck KGaA representative, the PD-L1 prototype assay has been sub-
jected to extensive testing of key performance characteristics. Analytical performance of
the PD-L1 IHC 73-10 assay demonstrated sensitivity and specificity for both the antibody
and the assay, within laboratory precision, working stability, and robustness that met accept-
ability criteria and were in keeping with reports of other assays for PD-L1 expression in the
literature. The method is, therefore, considered suitable for detecting PD-L1 expression in
formalin-fixed, paraffin-embedded histologic specimens from patients with solid tumors
(Dr. Hans Juergen Grote, written communication, September 2016).

Expression of PD-L1 as detected by PD-L1 IHC 73-10 assay localizes to membranous and/
or cytoplasmic regions in positive cells. For the analysis of PD-L1 expression in tumor cells,
only the staining of the plasma membrane is scored. The staining can be very strong and
homogenous (Figure 1); some tumors show heterogenous expression with variable inten-
sities (Figure 2). Strong and moderate
staining is easily identified at low-
power magnification (x20 or x40), but
the examination at high-power mag-
nification (x200) might be required to
identify weak staining. The staining
intensity is evaluated using a standard
semiquantitative scale of 0 (nega-
tive), 1+ (weak), 2+ (moderate), and
3+ (strong). In addition, the percent-
age of positive tumor cells with each
degree of staining intensity is recorded.
However, this “histo” (H) score has not Figure 1. A solid adenocarcinoma sample exhibiting a
been used in the clinical trials. Rather, diffuse, strong, and uniform staining with the Dako PD-L1 IHC
PD-L1 IHC is considered positive 73-10 assay.
or negative as compared with a pre-
defined cut-off—membranous staining
of any intensity in at least 1% of tumor
cells—established during early clinical
trials of avelumab. Tumor-infiltrating
lymphocytes can also be positive, but
the assessment of their expression is
not included in the cut-off for clini-
cal trials (Gulley 2015, Verschraegen
2016).
The 73-10 assay is still under devel-
opment so a clinically relevant cut-off
for PD-L1 expression has not been
Figure 2. A squamous cell carcinoma sample showing heterog-
definitively determined. However, enous PD-L1 staining with the Dako PD-L1 IHC 73-10 assay. Tumor
recommendations for reporting infor- cells showing moderate (2+) membranous staining are seen at the
mation, which are similar to those for periphery of the solid nest, in association with inflammatory cells
with strong programmed cell death ligand-1 (PD-L1) expression
other assays, can be found in Box 1. (arrows). The center of the nest consists of tumor cells with weak
or negative expression.
PD-L1 73-10 ASSAY 81
Box 1. Suggested Information for Inclusion Interpretation Pitfalls

When Reporting Results from the Dako Although the assay is under development, the inter-
PD-L1 IHC 73-10 Assay pretation pitfalls of the PD-L1 IHC 73-10 assay are
General Information likely to be simiar to the other assays. It is likely
• Positive control results (Pass/Fail) that nonspecific background, edge artifacts, crush
• Negative control results (Pass/Fail) artifacts, necrosis, or poor fixation might affect
• Whether adequate tumor cells (≥100
cells) are present (Yes/No)
interpretation. Several studies with multiple PD-L1
• Tumor Proportion Score: clones have reported intratumoral and intertumoral
PD-L1 IHC 73-10 Result to Clinician* heterogeneity of PD-L1 expression that could result in
• Expression <1% ___ discrepant results between different specimen types
• Expression ≥1% ___ (i.e., resection vs. biopsy and the primary tumor vs.
Additional Information metastasis (Ilie 2016, Kim 2015, McLaughlin 2015,
• Presence of tumor-associated Uruga 2017). In addition, PD-L1 expression can be
immune cells
affected by chemotherapy and/or radiation therapy
• Other comments to the clinician (Hecht 2016, Sheng 2016). Although similar data are
* The cut off value with clinical significance has not not available for the 73-10 assay, studies to address
been determined. these issues in general are warranted.
Similar to other clones, potential false-
positive interpretations could be attributed to alveolar macrophages that often
exhibit strong (3+) membranous staining and/or to stromal elements (inflammatory
cells and endothelial cells) that can show various intensities of staining (Figure 3).
Necrotic tumor cells might show cytoplasmic and/or irregular membranous staining
(Figure 4). Comparison with hematoxylin and eosin staining morphology may be useful to
exclude such non-tumoral staining, particularly in small biopsy samples.
Predictive Significance
Because avelumab is still in clinical development, only limited data is publicly available.
Results of a nonrandomized, phase I study in the first line setting (Verschraegen 2016
Figure 3. An adenocarcinoma sample with moderate-to- Figure 4. A squamous cell carcinoma sample with necro-
strong staining of endothelial cells and inflammatory cells sis. Necrotic debris (arrow heads) shows cytoplasmic and/
within the stroma. The vast majority of tumor cells are or fragmented membranous staining with the Dako PD-L1
negative. IHC 73-10 assay, whereas viable tumor cells exhibit heteroge-
neous (negative to focally moderate) membranous staining.
and Jerusalem 2016) and those in the second line or later setting (Gulley 2017) have been
reported (Table 1). Ongoing clinical trials using avelumab are summarized in Table 2.
Table 1. Summary of Clinical Trials with Response Data

Median Median Overall
Subgroup ORR Progression-free Survival
(%) Survival (months) (months)
Study Clinical
Design, Treatment PD-L1 PD-L1 PD-L1 PD-L1 PD-L1 PD-L1 Trial ID
Drug Phase Line pos. neg. pos. neg. pos. neg. (Name) Reference
Non- 12.0 8.9

17/122 2/20
randomized, ≥2 (HR = 5.9 (HR = 4.6 Gulley 2017
(14%) (10%)
phase Ib 0.45) 0.64) NCT
01772004
Avelumab
Non- (JAVELIN
(Pfizer/ 7/35 0/10 Verschraegen
randomized, 1 11.6 6.0 NA NA Solid
Merck (20.0%) (0.0%) 2016
phase Ib Tumor)
Serono)
Non-
19/88 2/23 Jerusalem
randomized, 1 NA NA NA NA
(21.6%) (8.7%) 2016
phase I
ORR = objective response rate, PD-L1 = programmed cell death ligand-1, HR = hazard ratio, NA = not applicable.
Table 2. Summary of Other Ongoing Clinical Trials

Tumor Type and Treatment Clinical Trial
Drugs Study Design and Phase Stage Line ID (Name) Reference
Avelumab vs. Randomized, open-label, NSCLC stage IIIB/IV or >2 NCT02395172 Park 2015
Docetaxel multicenter, global, phase III recurrent (Javelin Lung
200)
Avelumab vs. Randomized, open-label, NSCLC stage IV or 1 NCT02576574 Reck 2016
Platinum-based multicenter, global, phase III recurrent, with PD-L1 (Javelin Lung
Chemotherapy expression 100)
Doublets
Avelumab + Nonrandomized, open-label, NSCLC advanced or > 2 for Group NCT02584634 —
Crizotinib (A) or multicenter, phase Ib/II metastatic, ALK A; any line for (Javelin Lung
PF-06463922 (B) negative (Group A) or Group B 101)
ALK positive (Group B)
Avelumab + Randomized, open-label, Solid tumor advanced > 2 for phase NCT02554812 Ribas 2016
PF-05082566 (A) or multicenter, phase Ib/II or metastatic NSCLC Ib; any line (Javelin Medley)
PF-04518600 (B) for phase II
NSCLC = non-small cell lung cancer, ID = identifier, PD-L1 = programmed cell death ligand-1, ALK = anaplastic lymphoma kinase.
Avelumab (MSB0010718C) is a product of Pfizer/Merck Serono. Docetaxel is a product of (Taxotere) Sanofi-Aventis and (Docefrez) Sun
Pharma. Crizotinib (Xalkori), PF-06463922 (lorlatinib), PF-05082566 (Utomilumab), and PF-04518600 are products of Pfizer, Inc.
Conclusion
Although avelumab is in development, promising results similar to those reported for other
assays have been shown. It is expected that membranous staining of PD-L1 in at least 1%
of tumor cells will account for positive IHC. Reactions in immune cells are likely to be
excluded from evaluation.
9
Other Anti–PD-L1 Clones: Alternative
Assays and Laboratory-Developed Tests
By Lynette M. Sholl, Mari Mino-Kenudson, Reinhard Buettner, and Ignacio Wistuba
The US Food and Drug Administration (FDA) has approved a variety of companion and
complementary diagnostics for programmed cell death ligand-1 (PD-L1) expression testing
to help determine an appropriate PD-1/PD-L1 axis blockade therapy for a variety of cancer
types. The proliferation of these diagnostic assays poses special challenges for pathology
laboratories because most laboratories do not use all of the staining platforms required by
the different assay manufacturers, and use of companion diagnostic kits often increases the
cost of each individual test. Therefore, laboratories seeking to offer PD-L1 testing services
face significant capital and operating expenditures to acquire the equipment and reagents
necessary to offer a full range of companion and complementary PD-L1 diagnostics. Before
the approval of pembrolizumab by the FDA for patients with advanced non-small cell lung
cancer (NSCLC) in the first-line setting, many laboratories were sending out selected speci-
mens to commercial pathology laboratories that offered companion and complimentary
diagnostic kits, with the results incorporated in the pathology report when they became
available. The National Comprehensive Cancer Network guidelines implemented a rec-
ommendation post-approval that immunohistochemistry (IHC) testing for PD-L1 with a
validated assay should be performed for both advanced squamous cell and non-squamous
cell NSCLCs. The cost and administrative burden of sending tissue to reference laboratories
for PD-L1 testing may be prohibitive for many laboratories, given the number of patients
with advanced NSCLC seen in daily practice (especially in referral centers).
Lower-cost, laboratory-developed tests (LDTs) have been commercially available for
more than a decade, beginning with the mouse monoclonal anti-human PD-L1 antibody
29E.2A3 (Latchman 2001). However, this and other commercially available antibodies, such
as rabbit polyclonal anti–PD-L1 ab58810 (Abcam) and mouse monoclonal MIH1 (Thermo
Fisher Scientific), were shown to have lower specificity relative to another mouse monoclo-
nal antibody, 5H1 (Dong 2002, Velcheti 2014), developed and made available through the
Lieping Chen laboratory (Yale School of Medicine, New Haven, Conneticut) but not through
commercial vendors. This 5H1 antibody was used in the phase I trial of nivolumab, and a
correlation between tumor PD-L1 expression and response was demonstrated (Topalian
2012). In subsequent clinical trials of nivolumab, a novel and proprietary rabbit monoclonal
28-8 antibody (Agilent Technologies/Dako) replaced 5H1 and was subsequently incorpo-
rated into a complementary diagnostic kit (Phillips 2015).
Biomarker PD-L1 Antibodies

PD-L1 is a transmembrane protein, most of which is extracellular (including the PD-1 bind-
ing domain), with a 31 amino acid cytoplasmic domain. The 5H1, 7G11, 015, 22C3, and 28-8
antibodies bind the extracellular domain, whereas SP142, SP263, E1L3N, and 9A11 bind the
intracellular domain (Table 1). Although both membranous and cytoplasmic expression can
be seen in an antibody-dependent fashion, membranous staining has been correlated with
response in most clinical trials of PD-1 inhibitors. The mechanism leading to cytoplasmic
staining is not clear; however, it might represent accumulation of PD-L1 splice variants that
are not effectively localized to the membrane (Mahoney 2015).
Table 1. Anti–PD-L1 Antibodies for Use in Formalin-fixed Paraffin-embedded IHC

Antibody Validated for
Source Epitope Host
Clone Specificity
Lieping Chen laboratory, Yale Extracellular Dong 2002
5H1 Mouse monoclonal
School of Medicine domain Velcheti 2014
Freeman laboratory, Dana-
Extracellular
7G11 Farber Cancer Institute, Harvard Mouse IgG1 Mahoney 2015
domain IgV
Medical School
Extracellular
015 Sino Biological Rabbit IgG Mahoney 2015
domain
Extracellular Dolled-Filhart
22C3 Dako Mouse monoclonal
domain 2016
Extracellular
28-8 Abcam Rabbit monoclonal Cogswell 2017
domain
Cytoplasmic
SP142 Spring Bioscience/Ventana Rabbit monoclonal Mahoney 2015
domain
Cytoplasmic
SP263 Spring Bioscience/Ventana Rabbit monoclonal Smith 2016
domain
Cytoplasmic
E1L3N Cell Signaling Technology Rabbit monoclonal Mahoney 2015
domain
Cytoplasmic
9A11 Cell Signaling Technology Mouse IgG1 Mahoney 2015
domain
Ig = immunoglobin.
E1L3N Antibody
One of the most commonly used commercially-available antibodies used in LDTs is E1L3N
rabbit monoclonal antibody by Cell Signaling Technology, introduced in 2014. The preva-
lence of PD-L1 positivity using the E1L3N antibody in lung cancers ranges from 22% to
66%, depending on histology, platform, detection systems, and cut-off definition (i.e., 1%,
5%, or 50%, Table 2). In lung cancers, E1L3N has the highest sensitivity for membranous
expression when compared with SP142, 9A11, 015, and 7G11, with membranous expression
by the latter two extracellular-domain antibodies obscured by high cytoplasmic staining
OTHER ANTI–PD-L1 CLONES: ALTERNATIVE ASSAYS AND LABORATORY-DEVELOPED TESTS 85
Table 2. E1L3N IHC Conditions and Results in NSCLC

Dilution, Positive PD-L1
Antigen Platform/ Tumor
Reference Incubation Cut-off, Patients Positive
Retrieval Reagents Type
Time Pattern (%)
Inaguma 1:200, 30 High pH BOND-MAX 5% tumor and LUSC 56 58.9
2016 minutes Automated IHC/ immune cells, NS LUAD 137 21.9
ISH Stainer
Smith 17.5 μg/mL, Cell Benchmark Any NSCLC 100 24
2016 16 minutes Conditioning Ultra with
1 buffer x 64 Optiview
minutes detection
Igarashi 1:200, O/N Sodium SignalStain 1%, 5%, 10%, LUAD 150 92, 82,
2016 citrate buffer Boost IHC and/or 50%, 74, 48
pH 6.0 Detection cytoplasmic +
Reagent membranous
Paulsen 1:25, 32 Cell UtraMAP Intensity score LUSC 275 22

2016 minutes Conditioning HRP+ >1.25 (at least
1 buffer x 64 ChromoMAP weak to LUAD 503 24
minutes DAB moderate),
cytoplasmic and
membranous
Koh NS NS Benchmark XT ≥ 5%, LUAD 497 58

2015 Autostainer membranous
Scheel NS NS BOND-MAX Various cut-offs, LUSC 4 NA*

2016 Automated IHC/ membranous LUAD 11 NA*
ISH Stainer
Tang 2015 1:200, O/N Sodium citrate HRP-DAB H score ≥ 5%, NSCLC 170 65.9
buffer pH 7.4 cytoplasmic and
membranous
Huynh 1:200, O/N NS BOND RX ≥ 1%, LUAD 261 49

2016 Research IHC membranous
and ISH
Staining
Sheffield 1:200, NS HIER in Dako Auto- H score ≥ 1%, Non- 80 24

2016 DaVinci Green stainer Link 48 membranous squamous
Diluent x 35 platform NSCLC
minutes
Uruga 1:200 EDTA buffer BOND RX ≥ 1% Stages 109 51
2016 pH 9.0 membranous II and III
LUAD
Ameratunga 7 μg/mL, NS TRS pH9 + NS < 5%, LUSC 288 65.9

2016 microwave membranous at LUAD 182 53.7
intensity ≥ 2
Table 2 continued from previous page
Parra 1:100 Citrate buffer/ BOND-MAX Image analysis LUSC 34 31.5

2016 Tris-EDTA assessment,
buffer solution membranous, LUAD 34 23.3
various cut-offs
> 5%
Tsao 10.8 μg/mL, Cell Benchmark XT ≥ 1%, ≥ 25%, NSCLC 982 32, 20.8,
2017 60 minutes Conditioning ≥ 50% 14.3
1 x 90
minutes
* Interobserver concordance study using 1%, 5%, 10%, 25%, and 50% cut-offs. Programmed cell death ligand-1 = PD-L1, NS = not
specified, LUSC = lung squamous cell carcinoma, LUAD = lung adenocarcinoma, O/N = overnight, HRP = horseradish peroxidase,
DAB = 3, 3’diaminobenzidine, H = histo, NSCLC = non-small cell lung carcinoma, HIER = heat-induced epitope retrieval, TRS =
target retrieval solution. The Benchmark ULTRA and XT, Cell Conditioning 1 Solution (CC1), and UtraMAP HRP+ChromoMAP
DAB are products of Ventana Medical Systems, Inc. BOND-MAX Automated IHC/ISH Stainer and BOND RX Research IHC and ISH
Staining are products of Leica Biosystems. SignalStain Boost IHC Detection Reagent is a product of Cell Signaling Technology,
Inc. The Dako Autostainer Link 48 platform and Target Retrieval Solution (TRS) pH9 are products of Agilent Technologies/Dako.
DaVinci Green Diluent is a product of BioCare Medical. Tris-EDTA buffer solution is a product of Thermo Fisher Scientific.
(Mahoney 2015). The specificity of E1L3N for PD-L1 membranous expression was dem-
onstrated by its absence on cell lines that were engineered to have premature truncation
of PD-L1 (Cogswell 2017). A minority of PD-L1–deleted cells, however, continued to show
cytoplasmic expression with E1L3N but not with the 28-8 complementary diagnostic kit.
When compared with the 28-8 antibody using identical conditions, E1L3N showed intrinsic
sensitivity that is similar to or slightly higher as that for head and neck tumor cells, tonsil-
lar crypt epithelium, and immune cells (Cogswell 2017). Inaguma and colleagues compared
E1L3N antibody with the 28-8 antibody (Abcam) at 1:500 dilution using the BOND-MAX
Automated IHC/ISH Stainer (Leica Biosystems) in more than 5,000 tumor and normal
tissue samples (Inaguma 2016). Similar patterns of staining with the two antibodies were
observed; however, researchers ultimately favored E1L3N, due to decreased nonspecific
background staining.
In genetically engineered cell lines, E1L3N expression levels are highly concordant with
the antibodies SP142, 9A11, and SP263 using chromogenic IHC and quantitative immuno-
fluorescence (Gaule 2016). The concordance is decreased in studies of lung tumor tissue;
strikingly E1L3N and SP142 were discordant more than 25% of the time when identical
fields were examined by quantitative immunofluorescence for each antibody (McLaughlin
2015). This observation partially is attributed to tumoral heterogeneity because PD-L1
staining can show remarkable variability within a single tumor specimen (Gaule 2016).
However, comparative studies of diagnostic assays consistently show that the VENTANA
PD-L1 (SP142) Assay (Ventana Medical Systems, Inc.) stains fewer tumor cells when com-
pared with assays based on 28-8, 22C3, and SP263. In the National Comprehensive Cancer
Network comparison study, the particular E1L3N LDT used did show good concordance
with some commercial assays (see Chapter 11 for details). However, in the absence of assay
standardization, results using this LDT cannot necessarily be generalized.
Other Commercially Available PD-L1 Antibodies

A variety of anti–PD-L1 antibodies have been made commercially available during the past
several years. Some, such as the PD-L1 rabbit polyclonal CD274 antibody from Proteintech
Group, Inc., have been used in a variety of published studies across tumor types (Table 3,
Yang 2014, Yang 2016, Song 2016). Head-to-head studies of LDTs using antibodies other than
E1L3N are limited, however, Sheffield et al. demonstrated high agreement (Cohen’s kappa
of 0.69) among three LDTs (Tables 2 and 3) and the 28-8 companion diagnostic in a cohort
of non-squamous NSCLC samples. Correlation with RNA expression levels was very good
to excellent, as shown by in situ hybridization (Sheffield 2016). In a study by Paulsen et al.,
E1L3N was preferred over PD-L1 antibodies MAB1561 (mouse monoclonal, R&D Systems
Inc.) and ab58810 (rabbit polyclonal, Abcam) following evaluation using PD-L1–transfected
cell lines, with negative (brain) and positive (placenta) tissue controls (Table 2, Paulsen 2017,
Adam 2017). Studies of the 22C3 antibody using the companion diagnostic (PD-L1 IHC 22C3
pharmDx, Agilent Technologies/Dako) assay as compared with modified approaches using
two different Ventana detection systems demonstrated comparable, although not identical,
performance across assays (Neuman 2016).
Apart from antibody performance, the success and reproducibility of IHC-based assays
are heavily dependent on pre-analytic tissue handling (Gown 2016), as well as on the specific
characteristics of the antigen-retrieval and detection systems used. For PD-L1, the antigen-
retrieval conditions using citrate buffer pH6, citrate buffer pH 8, or EDTA buffer have been
shown to significantly affect the rate of positive PD-L1 expression for E1L3N (unpublished
observation). Available detection systems generate different levels of expression depending
on the strength of the amplification step (Figure 1). Amplification strength has the potential
to radically alter the outcome of the test and could lead to alternative PD-L1 expression
scoring when the expression level is near a cut-off threshold (see Chapter 3 for details).
Future Directions in Standardization of PD-L1 IHC Assays

All companion and complementary PD-L1 diagnostics were developed using proprietary
antibodies, many of which were made commercially available only after FDA approval. This
arrangement has hampered efforts to cross-compare the performance of FDA-approved in
vitro diagnostics and LDTs, however, data are now emerging on the comparative perfor-
mance of different antibodies and platforms in NSCLC (Chapter 11). The studies cited here,
and others under way, indicate that most antibodies, including those used exclusively in the
LDT setting, demonstrate comparable performance in well-controlled settings. Ultimately,
the interpretation is limited by variable definitions of positivity for each companion or com-
plementary diagnostic and by the lack of a clear gold-standard comparator, apart from the
commercial kits themselves. The relative lack of well-controlled outcomes data for patients
selected using an LDT and our limited knowledge of other factors that might modify response
to PD-1/PD-L1 axis blockade confound the use of PD-L1 positive or negative expression in
treatment selection, although clinical outcomes will inform this approach.
Another challenge associated with the use of LDTs as predictive markers in oncology
practice relates to the lack of clear oversight and dearth of standardization. In the United
States, proficiency testing programs, although commonplace for molecular diagnostics
laboratories and clinical pathology specialties, are not yet well established for IHC labs.
The College of American Pathologists (CAP) has developed a set of recommendations and
expert consensus opinions relating to the validation of IHC LDTs. The requirements are
Table 3. Other PD-L1 Laboratory-developed Tests, IHC Conditions, and Results in NSCLC
PD-L1
Dilution, Antigen Detection Tumor Expression
Reference Antibody Incubation Retrieval System Type Cut-off (%)
Takada PD-L1 (SP142) 1:100, O/N TRS (Dako) DAKO LUAD 1% 34.5
2016 (rabbit) 110°C x EnVision ≥ 5% 20.4
15 minutes FLEX
Yang CY PD-L1/CD274 1:250, 1 hour Citrate buffer UltraVision LUAD ≥ 5% 39.9
2014 (rabbit) 121°C Quanto
Detection
System HRP
DAB
Yang CY PD-L1/CD274 1:500, 1 hour Citrate buffer UltraVision LUSC ≥ 5% 56.2
2016 (rabbit) 121°C Quanto
Detection
System HRP
DAB
Zhang Y SAB2900365 1:300 Citrate buffer, NS LUAD Quickscore 49
2014 (rabbit) microwave > 8*
Song Z PD-L1/CD274 1:100, O/N NS DISCOVERY LUAD ≥ 5% 48.3
2016 (rabbit) CHROMOMap
DAB Kit (RUO)
Azuma PD-L1 NR, 30 Cell ultraView NSCLC H score 50

2014 (Lifespan minutes Conditioning Universal DAB = 30
Biosciences, 1 Solution Detection Kit
Seattle, WA) Retrieval
Sheffield PD-L1 (SP142) 1:100 HIER in (DaVinci Dako Non- 80 36

2016 (rabbit) Green Diluent ) Autostainer squamous
x 35 minutes Link 48 NSCLC
platform
Sheffield PD-L1/CD274 1:50 Tris-EDTA Dako Non- 80 38

et al 2016 Clone: RBT- buffer solution Autostainer squamous
PD-L1 (rabbit) (ThermoFisher Link 48 NSCLC
Scientific) x 30 platform
minutes
*This score equates to a minimum of 5% of tumor cells that stain with at least intermediate intensity. PD-L1 = programmed cell
death ligand-1, IHC = immunohistochemistry, NSCLC = non-small cell lung cancer, O/N = overnight, TRS = target retrieval solu-
tion, LUAD = lung adenocarcinoma, HRP = horseradish peroxidase, DAB = 3, 3’diaminobenzidine, LUSC = lung squamous cell
carcinoma,NS = not specified, NR = not reported, H = histo, HIER = heat-induced epitope retrieval.
The PD-L1 (SP142) rabbit monoclonal antibody cell line is available through Spring Biosciences and Ventana Medical Systems
Inc. The PD-L1/CD274 rabbit polyclonal antibody cell line is available through Proteintech Group Inc. The SAB2900365 rabbit
polyclonal antibody cell line is available through Sigma-Aldrich. The PD-L1/CD274 Clone: RBT-PDL-1 rabbit monoclonal antibody
cell line is available through Bio SB.
The DAKO EnVision FLEX and the Dako Autostainer Link 48 platform are products of Agilent Technologies/Dako. The UltraVision
Quanto Detection System HRP DAB and the Tris-EDTA buffer solution are products of Thermo Fisher Scientific. The DISCOVERY
ChromoMap DAB Kit (RUO), the ultraView Universal DAB Detection Kit, and Cell Conditioning 1 Solution are products of Ventana
Medical Systems Inc. DaVinci Green Diluent is a product of BioCare Medical.
Figure 1. Programmed cell death ligand-1 (PD-L1) staining of serial sections of a lung adenocarcinoma core
biopsy using the E1L3N antibody. The extent of staining varies depending on the antigen retrieval method and
the detection system. A) EDTA pH9 buffer solution was used as antigen retrieval and the Envision+ platform for
detection, with 20% tumor-cell positivity. B) A citrate buffer and Envision+ platform, with 20% tumor-cell posi-
tivity. C) A citrate buffer and the Envision FLEX platform, with 90% tumor-cell positivity but some cytoplasmic
PD-L1 expression. D) A citrate buffer and the Novolink Polymer Detection System (Leica Biosystems, Wetzlar,
Germany), with 90% tumor-cell positivity but some cytoplasmic PD-L1 expression.
most stringent for predictive markers (including PD-L1), recommending examination of 20

positive and 20 negative samples (Lin 2014). In practice, however, the CAP recommenda-
tion might be difficult to fulfill in the case of rare alterations or in situations where a gold
standard is not fully defined or readily available. The variable definitions of positive and
negative PD-L1 staining can also complicate this validation process for individual labs, but
might be mitigated by the performance of multiple cut-off comparisons. This clearly adds
up to a major validation exercise for any laboratory choosing to use an LDT.
NordiQC is an academic proficiency testing program focused on external quality assess-
ment that initially involved Nordic-country laboratories but has since expanded to more
than 700 laboratories in 80 countries. In this scheme, tumor microarrays of formalin-fixed
paraffin-embedded materials are distributed to participating laboratories for staining and
returned for central expert assessment of qualitative variables including staining inten-
sity, signal to noise, and morphology. Participants receive information on recommended
antibodies and protocols, as well as tailored feedback for improving insufficient protocols.
According to the NordiQC data, false negativity is a common reason for insufficient results
and can be mitigated by improved epitope retrieval or enhanced-sensitivity detection sys-
tems (Vyberg 2016). Standardization programs established for predictive IHC markers,
such as for ALK expression in lung cancer, have successfully led to high inter-laboratory
concordance (Cutz 2014), followed by the development of national proficiency testing in
Canada for ALK IHC and fluorescence in situ hybridization testing (Cheung 2015). The UK
National External Quality Assessment Service (UK NEQAS) center in the Royal Infirmary of
Edinburgh, Edinburgh, United Kingdom, is in the process of developing an external quality-
assessment scheme for PD-L1 IHC. It is also worth noting that, according to UK NEQAS
external quality-assessment data for ALK expression IHC testing, laboratories using LDTs
had a significantly greater chance of failing an external quality assessment when compared
with laboratories using commercial kit assays (Ibrahim 2016).
Standardization of PD-L1 LDTs should involve head-to-head comparisons of PD-L1
expression levels determined by LDTs compared with levels determined by approved com-
panion diagnostics, such as the
PD-L1 IHC 22C3 pharmDx in the context of first line therapy, using an adequate number
of positive and negative samples (such as per CAP guidelines) and/or tissue microarrays. It
is recommended that a comparison of tumor proportion scores by the two methodologies
is performed to detect any systematic bias in the LDT relative to the companion diagnostic.
Due to the multiple factors that could influence the outcome of any IHC test (see Chapter 3
for details) and to the lack of any clinical outcomes data using anything other than a trial-
validated commercial assay, the only standard that can be used to gauge the likely clinical
predictive performance of an LDT would be a commercial assay.
Recommendations for PD-L1 Testing by LDTs: Validation and Performance

PD-L1 IHC using LDTs should be carried out on approximately 4-μ sections, cut fresh when-
ever possible. Immunoreactivity tends to degrade with time, and use of slides cut more than 6
months prior should be avoided, although the performance of available antibodies on archival
tissues is limited. Shorter time periods for storage of cut sections might be problematic if
sections are not stored in cool, dry, and dark conditions (Chapter 3). A gold standard posi-
tive control has not been formally proposed, however, acceptable positive control samples
include tonsil epithelium, placenta (placental trophoblasts highly express PD-L1, Inaguma
2016), and PD-L1 cell lines with high expression, such as NCI-H226 (lung large cell carci-
noma), NCI-H1975 (lung adenocarcinoma), or HDLM2 (Hodgkin lymphoma). Brain tissue
or known negative cell lines, such as H549 or PC3, may be used as negative controls. It is
recommended to use controls with different levels of PD-L1 expression including negative,
low, intermediate, and high levels. This may be facilitated by use of commercially available
cell lines engineered to express membranous PD-L1. These cell lines are developed for PD-L1
IHC controls that have both high and, more importantly, low levels of epitope expression
(see Chapter 3 for details). Use of tumors with known high levels of innate PD-L1 expression
(typically a result of CD274 amplification), such as Hodgkin lymphoma or human papilloma
virus-driven cervical squamous cell carcinoma may also serve as positive comparators in
a validation set. Igarashi et al. proposed the use of alveolar macrophages as an internal
control and intensity score reference for tumor staining in lung specimens (Igarashi 2016);
however, the intensity of staining in this cell type can vary considerably among individual
samples. Therefore, additional external positive control(s) showing consistent expression
levels—particularly including a low expression control—is recommended in routine practice.
If, however, PD-L1 expression is absent in both a tumor sample and the associated alveolar
macrophages of a lung biopsy sample, an effort should be made to confirm that a sample is
appropriately immunoreactive and is not falsely negative. Validation and proficiency also
may be facilitated by sample exchange among diagnostic laboratories, both to confirm accept-
able antibody/platform performance and to confirm good inter-pathologist agreement about
established cut-off levels, such as at 1% and 50% tumor proportion scores. Inter-pathologist
reproducibility for tumor cell scoring is high in multiple studies, but the same is not true for
quantification of PD-L1 expression on immune cells (Rimm 2016, Scheel 2016).
Conclusion
LDTs using biologically validated antibodies and clinically validated techniques may be an
acceptable and economical alternative to companion or complementary diagnostics in iden-
tifying tumors with PD-L1 overexpression—with caveats. In a multicenter study comparing
several commercial assays, as well as several LDTs using the same antibody clones as used
in the commercial assays, 50% of the LDTs did not have adequate comparative technical
performance (Adam 2017). The standardization of platform and staining conditions is of
critical importance for consistent and reproducible results. Ongoing studies to establish
equivalence to established companion or complementary diagnostics should provide support
for the use of stringently validated LDTs as predictors of response to approved immuno-
therapies. However, a more robust international proficiency testing infrastructure will be
essential to promote high-quality and consistent PD-L1 IHC results across antibodies and
test platforms and in a variety of settings.
10
Complementary and Companion
Diagnostics
By Fred R. Hirsch and Sanja Dacic
The commercial success of drugs such as tratuzumab and imatinib, which both require
companion diagnostics before they can be prescribed, has advanced the co-development
of therapeutic products and accompanying in vitro diagnostics. There are now numerous
examples of therapeutic products with an accompanying companion diagnostic (FDA. List
of Cleared or Approved Companion Diagnostic Devices).
With the introduction of personalized medicine, the US Food and Drug Administration
(FDA) introduced the term “companion diagnostics,” which is defined as a diagnostic
assay required for the use of the associated drug based on clinical efficacy and safety data.
There are several drugs approved by the FDA with companion diagnostics that assist with
therapeutic selection for treatment of patients with non-small cell lung cancer (NSCLC):
the Vysis ALK Break Apart FISH Probe Kit (Abbott Laboratories, Abbott Park, Illinois) or
the VENTANA ALK (D5F3) CDx Assay (Ventana Medical Systems, Inc., Tucson, Arizona)
for crizotinib, EGFR mutation testing with the Therascreen EGFR RGQ PCR Kit (Qiagen,
Hilden, Germany) for gefitinib and afatinib, and the Cobas 4800 System (Roche Molecular
Diagnostics Inc., Pleasanton, California) for erlotinib.
There are several reasons why there is so much interest in the development of companion
diagnostics. The main advantage of companion diagnostics is to segregate a patient popula-
tion into two subsets: biomarker positive and biomarker negative in order to ensure patients
have the highest chance of clinical benefit on a safety basis. This separation is based on a
quantitative assay result that is translated into a qualitative result, which represents a clinical
decision point, or cut-off. Furthermore, the safety and efficacy of the therapeutic product
is evaluated in the population that is treated in the clinical trial. Not all of the clinical trials
for drugs in development will be successful, and a companion diagnostic is one of the few
tools available to drug developers that can accelerate identification of the patient population
most likely to benefit from a specific therapeutic agent. In turn, this gives the therapeutic
agent a higher chance for achieving regulatory approval. Furthermore, enrollment of selec-
tive patients using a companion diagnostic can reduce the duration of the clinical trial. This
strategy resulted in a dramatic increase in biomarker-targeted drug-development programs.
In the early 1990s, 5% of new drug approvals were for targeted therapies, whereas this per-
centage increased to 45% in 2013.
In 2015, the FDA introduced the term “complementary diagnostics” for personalized
therapy, which is defined as a diagnostic assay that predicts a favorable outcome of the
associated drug by selecting patients based on results of the assay. However, it is not harm-
ful to treat patients with the associated drug in the absence of assay results or if the results
are negative. In other words, use of the complementary assay is not required for treatment
with the associated drug.
The differences between these two types of assays have been extensively reviewed
elsewhere (Milne 2015). Complimentary diagnostic test outcomes fall into the “nice to
know, but not required” category, which means that results are variably interpreted and
used by oncologists. Requiring use of a companion diagnostic assists with consistency of
test use, although it may also be viewed as inconvenient, depending in individual perspec-
tives. Laboratories have more leverage, however, if test funding is an issue but test use is
a requirement. Use of the test should result in a higher probability of treatment benefit
because the patient group has been better defined. However, inconsistent use of these tests
and variable interpretation of results allows increased patient access to treatment groups,
widening treatment populations. The flexibility may be appreciated by some, but this may
make it harder for some laboratories to get funding for a test that is not viewed as necessary.
Examples
The PD-L1 IHC 28-8 pharmDx (Dako, Glostrop, Denmark) was given complementary diag-
nostic test status based on the results of the CheckMate 017 clinical trial. In this trial, overall,
patients treated with nivolumab had longer survival than those treated with docetaxel,
although a subset analyses clearly suggested that the level of programmed cell death ligand-1
(PD-L1) expression in NSCLC tumors might help identify patients who are more likely to
benefit from nivolumab. Furthermore, those patients who had PD-L1 expression levels
less than the biomarker threshold of 1% did no worse with nivolumab when compared
with docetaxel. The results, therefore, showed clinical benefit of the treatment without
patient selection, but the benefit was greater for the selected patients. If use of nivolumab
was based on results of a required companion diagnostic, some patients would lose the
chance to receive a potentially beneficial treatment. Nivolumab has been approved by the
FDA for second-line therapy of patients with advanced NSCLC, with no requirement for
a companion diagnostic. The 28-8 pharmDx assay, however, serves as a complementary
diagnostic using a cut-off value of PD-L1 expression of at least 1% (see Chapters 2 and 4).
Pembrolizumab has been approved by the FDA and the European Medicines Agency for
both first-line and second-line therapy of patients with advanced NSCLC. The PD-L1 IHC
22C3 pharmDx (Dako, Glostrop, Denmark) assay is a companion diagnostic that defines
positive PD-L1 expression as expression by at least 50% of tumor cells in the first-line set-
ting, and 1% or greater of tumor cell expression for second-line therapy.
Some multi-industrial/academic collaborative studies, such as the “Blueprint Project”
(Chapter 11), have been published (Hirsch 2017) and others are ongoing, with the goal to
compare the analytical and diagnostic performance across the various assays for PD-L1
expression.
COMPLEMENTARY AND COMPANION DIAGNOSTICS 95
Validation Processes and Diagnostic Tests

Both companion and complimentary diagnostic assays undergo rigorous analytical validation
processes for accuracy, reproducibility, specificity, sensitivity, and stability. In addition, both
types of diagnostic assays require use of particular analytical reagents and instrumentation,
which makes implementation in diagnostic laboratories challenging (Kerr 2015, Kerr 2016a).
Each of the five drugs that are furthest in the development process for NSCLC, have their
own trial-validated assays (as discussed throughout this Atlas), which creates complexity
of choice. There are several diagnostic platforms for PD-L1 IHC, all of which use different
antibody clones, staining protocols, platforms, and, most importantly, different clinical deci-
sion points. Most diagnostic laboratories are not able to adopt and run multiple diagnostic
platforms due to cost effectiveness and to the increasingly important role of reimbursement
for therapeutic tests.
Health care reform in the United States is changing the model for clinical provider
payment, moving to outcomes-driven reimbursement compared with testing driven. The
patient-outcome payment model is based on clinical outcomes for patients as opposed to the
number of clinical interventions and diagnostic tests ordered and implemented by physi-
cians. The current development strategy of a single diagnostic test for a single therapeutic
agent is challenged by this new payment model because the one-for-one approach is unlikely
to predict outcome in a large patient population. Payers are becoming very interested in
diagnostic tests that provide information about multiple potential treatment options that
could be provided by multiplex assays rather than by a single assay. This approach opens the
question as to whether the same diagnostic test can be used for similar therapeutic agents,
such as in the case of immune check point inhibitors. This is discussed in more detail in
Chapter 11.
Conclusion
Use of a companion diagnostic assay is a requirement for drug eligibility and prescription to
ensure the highest chance of clinical benefit on a safety basis, whereas use of the comple-
mentary assay is optional, with results informing but not dictating treatment decisions. A
better understanding of the different PD-L1–expression assays hopefully will lead to a more
rational use of these tools in clinical practice.
Assay Harmonization: Is It Possible?
By Keith M. Kerr, Fred R. Hirsch, Yasushi Yatabe, and Ming S. Tsao 11
There are detailed descriptions of five different programmed cell death ligand-1 (PD-L1)
immunohistochemistry (IHC) assays elsewhere in this Atlas. Each of these assays has been
separately developed, or is undergoing validation, in association with a specific anti-PD-1/
PD-L1 drug (Table 1). The 28-8 clone-based assay (Agilent Technologies/Dako) is now reg-
istered with the US Food and Drug Administration (FDA) as a complementary diagnostic
in association with nivolumab, whereas the 22C3 assay (Agilent Technologies/Dako) is a
companion diagnostic for pembrolizumab. The SP142 assay (Ventana Medical Systems,
Inc.) is approved for use with atezolizumab as a complementary diagnostic. At the time of
this publication, the SP263 clone-based assay (Ventana Medical Systems, Inc.), associated
with durvalumab, is available for use in Europe but not in the United States, and a 73-10
antibody clone-based assay for use with avelumab is in development. Recently, based upon
some of the technical comparison studies discussed below, the SP263 assay has also become
available as a complementary test for use with nivolumab. Unlike the trial validated 28-8
assay, however, there are no clinical validation data relating to SP263 use with nivolumab.
The assays developed by Dako (Table 1) have been developed for use on a Dako automated
IHC staining platform (the Dako Autostainer Link 48), whereas the two Ventana assays are
being developed for use on a Ventana platform (the Benchmark XT, GX, or Ultra).
The majority of the emerging evidence, reviewed elsewhere in this Atlas, indicates
that the anti-PD-1/PD-L1 drugs (Table 1) all show superior response and survival rates
for tumors with PD-L1 IHC expression above a given detection threshold, when compared
with tumors with expression under the threshold. These tests—whether used as compan-
ion or complementary diagnostic assays—will be used, either by regulatory authorities or
through physician preference, to inform the prescription of these drugs. This will be the
case for use in all lines of treatment. Anti–PD-L1 IHC testing will be required in non-small
cell lung cancer (NSCLC) for the foreseeable future.
The rapid emergence of these five drug–assay combinations poses some unique challenges
for the pathology and the oncology communities (Kerr 2015; Kerr 2016a; Kerr 2016b; Sholl
2016). Some of the issues involved have been encountered before. The pathology community
has experience regarding companion diagnostics and the use of specific FDA-approved
assays, such as with HER2 testing in breast cancer for trastuzumab therapy and with ALK
IHC in NSCLC. The lung cancer oncology community is familiar with the use of diagnostic
assays to identify patients with EGFR mutations or ALK or ROS1 translocations for the selec-
tion of an associated tyrosine kinase inhibitor therapy. What is unique about the PD-L1 IHC
story is the emergence of five different drugs, each with its own specifically developed, trial
associated, and validated assay. It is crucial to understand that IHC is a rather different type
of assay method when compared with mutation testing or fluorescence in situ hybridization
(FISH) testing. Although the latter two techniques may involve numerous and quite diverse
methodologies (different probes in the case of FISH testing), there is a common factor in
the test—namely the abnormal DNA (or RNA) sequences being sought. It does not matter
how the abnormal DNA sequence is identified, provided the techniques are performed using
adequate laboratory standards and are quality assured. IHC, however, is different. Although
each of the five IHC assays identifies PD-L1 protein expression, each antibody clone will
be specific for a different part (epitope) of the PD-L1 protein. In addition, each antibody
clone will not necessarily have the same binding affinity for its epitope because of the way
individual clones of plasma cells (hybridomas) are developed to produce anti–PD-L1 anti-
bodies. Both the selection of a primary antibody clone and the detection method used to
generate the color stain (chromogen) on the slide are equally important to the outcome of
an IHC assay. Different assays may use different chemistry, with or without amplification,
to boost color generation on the slide (see Chapters 4 to 8 for details). Consequently, these
five assays are not necessarily the same; therefore, it is not a given that they will perform
is the same way.
Table 1. Five-Assay Comparison*
Clinical Cut-off(s) for

Assay Antibody Staining Immunotherapy PD-L1 FDA
PD-L1 Clone Platform Drug Expression Designation
28-8 Dako Link 48 Nivolumab ≥ 1%, ≥ 5% Complementary
(Bristol-Myers Squibb) device
22C3 Dako Link 48 Pembrolizumab ≥ 1%, ≥ 50% Companion

(Merck) device
SP142 Ventana Atezolizumab Tumor cells Complementary
Benchmark or (Genentech/Roche) ≥ 1%, ≥ 5%, ≥ 50% device
Ultra Immune cells
≥ 1%, ≥ 5%, ≥ 10% by area
SP263 Ventana Durvalumab ≥ 25% No designation

Benchmark or (AstraZeneca/
Ultra MedImmune)
73-10 Dako Link 48 Avelumab ≥ 1%, ≥ 50%, ≥ 80%, In development
(Pfizer/Merck Serono)
* Details for each assay appear in corresponding chapters (4 to 8). The 28-8, 22C3, and 73-10 assays, as well as their related
platform, are products of Agilent Technologies/Dako. The SP142 and SP263 assays, as well as their related platforms, are
products of Ventana Medical Systems, Inc.
ASSAY HARMONIZATION: IS IT POSSIBLE? 99
Five Drugs and Five Assays: Practical Problems

Many trials have used these agents in a range of different settings, and more trials are ongo-
ing. How these drugs might be used for different indications and in clinical settings is yet to
be determined. It is very likely, however, that for any given setting—stage of disease, lung
cancer subtype, single-drug or combination therapy, and various lines of therapy—there
will be several competing drugs available.
Each commercially produced and validated assay is designed for use on a particular
staining platform (Dolled-Filhart 2016; Novotny 2016), and laboratories tend to use only
one type of platform. The assay reagent packaging is usually only compatible with its
manufacturer’s platforms and not with competitors’ technology. Will laboratories be willing
or able to invest in alternative IHC-staining platforms for the delivery of a single NSCLC
diagnostic test, which might represent a very small volume of work per month? At least
some of the assays are being commercially developed for alternative staining platforms,
which might help laboratories better solve this problem.
Biomarker testing in NSCLC involves either pathologist-ordered reflex testing at diagno-
sis or custom testing, as requested by the oncologist or tumor board. If drug-specific assays
must be used, reflex testing will be difficult because the pathologist will not necessarily
know which drug the oncologist intends to use. Requests for custom testing would need to
include the intended therapeutic agent. This could be further complicated by the fact that
the treatment thresholds, or cut-offs, are different for some of the drugs; some tumors will
express PD-L1 below the threshold for treatment with one drug but above the threshold
for another.
Pathologists will require some form of training to effectively report the results of these
assays. Despite every pathologist’s best effort when reading IHC slides, experience tells us
that training is essential for consistent and reproducible reporting (Rüschoff 2013). The
reportable cut-offs are different for some assays and indications. For example, the SP142
assay used with atezolizumab determines PD-L1 expression on immune cells to reflect the
proportional area of tumor infiltrated by those cells if the proportion PD-L1 expression on
tumor cells is less than 1% (Fehrenbacher 2016; Chapter 6). Developing the expertise to
reproducibly deliver consistent results from PD-L1 IHC testing using different assays will
require a significant commitment by pathologists to become adequately trained in the use
of each assay.
Laboratory performance in IHC testing should be monitored by external quality-assur-
ance (EQA) schemes. To date, such EQA schemes focus on the IHC marker (protein) being
assessed. The laboratory performs the test according to its own practice, which could
involve use of a commercially available assay kit or a range of available primary antibody
clones that were raised against the same protein biomarker, as well as various methods of
antigen retrieval and detection, known as laboratory-developed tests (LDTs) (Chapter 9).
The presence of five specific assays that are individually associated with five drugs will
significantly complicate the development of PD-L1 IHC EQA schemes. Furthermore, EQA
schemes have traditionally measured the technical performance of the staining assay, not
the ability of the pathologist to correctly read the slide.
Demonstrating a Need for a Single PD-L1 IHC Test: The Role of the
Pathology Laboratory
Challenges for pathologists and oncologists regarding the availability of up to five different
drug–assay combinations for a given indication in NSCLC have been described previously in
this chapter and elsewhere (Kerr 2015; Kerr 2016a; Kerr 2016b; Sholl 2016). The biology of
each PD-L1 IHC assay is different. The outcomes for patients, when using these drugs in a
population selected for greater levels of PD-L1 expression, have only been validated in trials
using the specific drug–assay combinations listed in Table 1. The practical challenges of
providing up to five different PD-L1 IHC tests to support the use of five drugs in any institu-
tion have, however, led to the inevitable question: Can we use only one of these assays—or
any other LDT for that matter—to select patients for any anti–PD-1/PD-L1 therapy?
Pathology is slowly and, to some extent, reluctantly coming to terms with commercially
produced IHC kit assays. These kits are generally relatively expensive, far exceeding the cost
of an LDT developed using individually sourced components. Traditionally, all diagnostic
IHC tests have been developed as LDTs, as antibody clones to an ever-expanding range of
biomarkers became available. Experience with EQA schemes that assess the performance
of laboratories in routine diagnostic IHC performance has demonstrated a wide variation
in performance (http://www.ukneqas.org.uk). Although IHC in the traditional diagnostic
setting is an adjunct to morphology-based section diagnosis using hematoxylin and eosin
staining, it is important that the IHC is performed to an adequate standard. When labo-
ratories introduce a new IHC test to their repertoire, CAP has rigorous recommendations
for the technical validation of the test based on a large number of test cases (Fitzgibbons
2014; Lin 2014). Anecdotal evidence suggests that not all laboratories adhere to these rec-
ommendations, however. The stakes for the patient are higher regarding use of companion
or complementary IHC assays because the IHC test outcome is no longer just an adjunct
to diagnosis—it is the core, treatment-determining metric. In this scenario, consistency
and accuracy are absolutely vital, not only to guarantee correct technical performance of
the assay but also to ensure that the clinical outcome for the patient, as predicted from the
clinical trial, can be reproduced in the treatment setting. Experience from the ALK IHC
EQA, run by the UK National External Quality Assessment Service, has shown that adequate
technical performance in ALK IHC testing is more likely to be achieved with a commercial
ALK IHC kit compared with an LDT (Ibrahim 2016).
If a single PD-L1 IHC assay is to be used for the selection of all available associated
drugs, how should a laboratory decide which assay to use? Are all commercially developed,
trial-validated assays the same? How technically comparable would an LDT developed from
available reagents actually be? These questions can be answered by comparative studies
assessing the technical performance of assays using a set of NSCLC tumor samples, all
stained by the assays to be compared. These studies would allow for comparison of staining
outcomes based on staining intensity and distribution in the same sample. It would also be
possible to assess any variance in the allocation of expression levels above or below clinical
thresholds for treatment selection. However, these studies would not tell us whether the
expected probabilities of response to treatment and survival outcomes, predicted from trials
for the drug-associated assays (Table 1), can be reproduced when alternative tests or assays
are used to select patients for treatment. Working toward this goal, the only gold standard
we have (in the absence of validating clinical outcome data) is comparison with at least one
of the trial-validated commercial assays.
Demonstrating Concordance Among Assays: Comparative Studies

There is a growing literature base on this subject, which is not surprising because three of
the commercially produced and trial-validated assays have only become available in the
past year. A few studies have been published using various LDTs and methods for slide
assessment (Velcheti 2014; Sheffield 2016; McLaughlin 2016), but these are of limited value
because they do not involve the trial-validated assays and, as mentioned previously, there is
no obvious gold standard for assessment of test performance. Unsurprisingly, these studies
demonstrated that the assays being compared were not the same. In the absence of any data
comparing clinical outcomes in patients selected for a treatment using alternative PD-L1
IHC assays, comparison of alternative assay performance with the performance of one or
more of the trial-validated assays is the only valid method of comparison. One recent study
showed that the differences observed between commercial assays and LDTs, using quan-
titative IHC, may be a result of the associated chemistry used to build the assays and not a
result of the differences between the individual primary antibodies SP142, E1L3N, 9A11,
SP263, 22C3, and 28-8 themselves (Gaule 2016).
Scheel et al. recently published data from a ring study analyzing interobserver agree-
ment of PD-L1 IHC scoring and comparing assay performance (Scheel 2016). This study
used surgically resected samples from 15 patients with lung cancer, stained with four of
the assays listed in Table 1. Sections of each tumor were stained using the SP142 and SP263
assays at a Ventana laboratory. The 22C3 assay staining was performed at a Dako labora-
tory, and the 28-8 assay staining was performed at an independent university laboratory
in Germany. Each pathologist allocated each patient a score based on the percentage of
tumor- and immune-cell staining, as defined by six ranges that were determined by several
cut-offs that have been used in clinical trials. The six categories were: less than 1%, 1% to
less than 5%, 5% to less than 10%, 10% to less than 25%, 25% to less than 50%, and 50% or
greater. There was moderate concordance between observers when allocation was compared
across the six different categories (K = 0.47-0.49). Within these comparisons, approximately
57% to 60% of pairs were concordant, 25% to 32% were discordant for one category, and
10% to 15% were discordant for two categories. Concordance was poor for the assessment
of immune-cell staining using the six-category approach (K < 0.2). As expected, concor-
dance for tumor-cell staining was better if only two categories were considered, above or
below various thresholds (k = 0.59 – 0.80). There was no difference in concordance among
assays. Comparison of category allocation across assays showed that use of the 28-8 and
22C3 assays resulted in similar stained populations tumor cells in 12 out of 15 cases, and a
72% concordance was seen for allocation to the six ranges. The SP142 assay stained fewer
tumor cells in four cases, whereas SP263 stained more tumor cells in nine cases. This led
to a 53% to 56% concordance for allocation to the six categories when comparing the SP142
or SP263 assays with the Dako assays. Only 41% concordance was demonstrated, however,
when the SP142 and SP263 assays were compared with one another.
Ratcliffe et al (Ratcliffe 2017) presented a comparative study of three commercially
available, trial-validated assays based on 28-8, 22C3, and SP263 antibodies. The study used
500 commercially sourced NSCLC tumor specimens, which were stained in a commercial
laboratory using Clinical Laboratory Improvement Amendments guidelines. The specimens
were read by a single pathologist from the same laboratory. The stains were all scored for the
percentage of tumor cells that showed PD-L1 expression, and pairwise comparisons were
made between pairs of assays. This study showed that the technical performance of these
three assays was very similar, with greater than 90% overall agreement in all comparisons
across the total range of PD-L1 expression. One possible source of bias in favor of concor-
dance in this study is the large proportion of specimens with completely negative staining.
In conjunction with Bristol-Myers Squibb, the National Clinical Cancer Network con-
ducted a comparative study using surgically resected samples from 90 patients with NSCLC.
Samples were stained using three of the trial-validated assays (28-8, 22C3, and SP142) and
an LDT developed using the E1L3N antibody clone (Rimm 2017). The staining was per-
formed in an academic laboratory environment. Comparisons were made relating to overall
expression in tumor and immune cells, and to the determination of expression levels as
above and below several thresholds. The study concluded that the 28-8 and 22C3 assays
and the E1L3N LDT were all very analytically similar for tumor-cell staining (concordance
= 0.813), whereas the SP142 assay stained fewer tumor cells.
At the IASLC World Congress on Lung Cancer in December 2016, Adam et al. presented
data from a French PD-L1 IHC harmonization study (Adam J et al, WCLC 2016). This study
involved 41 surgically resected tumors, which were stained using the 28-8, 22C3, SP142,
and SP263 commercial assays, as well as LDTs developed in several academic laboratories
and based on the same four antibody clones or on the E1L3N clone. Once again, the 28-8,
22C3, and SP263 commercial assays showed concordant technical performance. Notably,
there was 95% concordance in tumor assignment using the 50% cut-off. It is also telling
from this complex study that 50% of the LDTs used in this study showed poor correlation
with trial-validated, commercially developed assays.
There are other comparison studies worth noting. Two alternative assays using the 22C3
antibody clone and the Ventana Benchmark XT staining platform showed approximately
85% concordance with the Dako trial-validated assay (Neuman 2016). This study used the
primary antibody reagents provided in the 22C3 commercial assay kit. A separate study
examined 219 surgically resected adenocarcinomas in a tissue microarray that were stained
using the 22C3 and SP263 commercial assays, as well as an SP142 clone-based LDT. The
22C3 assay and the SP142 clone-based LDT showed similar results, with 94% concordance;
however, the SP263 assay showed greater levels of staining, which led to decreased concor-
dance with the other two tests (74% to 76.3%) (Yeh YC ESMO Asia).
The IASLC BluePrint Study

The BluePrint study is a collaboration among four pharmaceutical industry partners (Roche/
Genentech, Astra Zeneca, Bristol-Myers Squibb Company, and Merck), two diagnostic part-
ners (Ventana Medical Systems, Inc. and Agilent Technologies/Dako), and the International
Association for the Study of Lung Cancer (IASLC). The purpose of this study was to compare
staining for the four commercially available trial-validated assays using the same NSCLC
tumor samples. In phase I, 40 commercially sourced NSCLC tumor samples were selected
from a much larger cohort to reflect the full dynamic range of PD-L1 expression. Sections
of each tumor sample were stained at either a Ventana laboratory (for the SP142 and SP263
assays) or a Dako laboratory (for the 28-8 and 22C3 assays). The cases were read by in-house
pathologists who were expertly trained in their company’s assays. The raw percentage of
tumor and immune cells stained by each assay was determined, and cut-offs for each assay,
as dictated by the manufacturers upon FDA approval, were used. PD-L1 expression cut-offs
were as follows: greater than/less than 1% of tumor cell staining for the 28-8 and 22C3 assays,
greater than/less than 25% of staining for the SP263 assay, tumor- and immune-cell scoring
of TC0-3/IC0-3 for the SP142 assay, and tumor- and immune-cell scoring of TC1/IC1 for
the SP142 assay (see Chapter 6 for tumor- and immune-cell scoring definitions) (Hirsch 2017).
PD-L1 expression for each tumor sample was determined to be above or below the
selected threshold for treatment determination when each assay was read using its own
algorithm (Figure 1). Comparison was then made between assays, with each of the treatment-
determining cut-offs applied to each of the assays (Table 2). It is important to note that this
study did not involve any immunotherapy selection and treatment; it merely examined
whether PD-L1 expression for individual patients would have been determined as above or
below various detection thresholds. These data could potentially be used to select patients
for immunotherapy.
30/38 26/38 26/38 20/38 Number of samples (%) with programmed

samples samples samples samples cell death ligand-1 (PD-L1) expression
(78.9%) (60.5%) (60.5%) (52.6%) above the assay specific selected threshold.
38
37
36
35
34
33
32
31
30
29
28
27
Increasing PD-L1 Expression
26
25
24
23
22 N=19 concordant above threshold
21
Case Number
20
19
N=14 discordant above threshold
18
17 N=5 concordant below threshold
16
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
SP142 22C3 28-8 SP263

TC1/IC1 1% TPS 1% TPS 25% TPS
PD-L1 expression below the selected cut-off
PD-L1 expression above the selected cut-off
Figure 1. Comparison of sample allocation above or below various thresholds for clinical assays. Reproduced with permission
from Hirsch FR et al, J Thorac Oncol. 2017;12(2):208-222..
Table 2. Concordance of PD-L1 Status Among Assays When Specific Clinical Cut-offs Were Applied
Scoring Algorithm Used
SP142/TC1/
PD-L1 22C3/1% Cut-off 28-8/1% Cut-off IC1 Definition* SP263/25% Cut-off
Antibody (Samples Used/ (Samples Used/ (Samples Used/ (Samples Used/
Clone Base Total Samples, Total Samples, Total Samples, Total Samples,
for Assay % Concordant) % Concordant) % Concordant) % Concordant)
22C3 38/38 (100%) 36/38 (94.7%) 33/38 (86.8%) 34/38 (89.5%)

28-8 36/38 (94.7%) 38/38 (100%) 31/38 (81.6%) 33/38 (86.8%)
SP142 24/38 (63.2%) 24/38 (63.2%) 38/38 (100%) 25/38 (65.8%)
SP263 34/38 (89.5%) 34/38 (89.5%) 33/38 (86.8%) 38/38 (100%)
* Tumor cell (TC) and immune cell (IC) scoring ranges are described in chapter 6. TC0 is defined as less than 1% of tumor cells
expressing PD-L1, TC1 is 1% to 5% expression, TC2 is 5% to 50% expression, and TC3 is greater than 50% expression. Table
adapted from Hirsch FR et al, J Thorac Oncol. 2017;12(2):208-222.
Across the dynamic range of staining produced by these assays, the 28-8, 22C3, and
SP263 assays were shown to be very similar in performance in the same set of NSCLC tumor
samples. The SP142 assay, however, stained consistently fewer tumor cells (Figure 2). These
findings match those found in the other studies reported here. Although there was no outlier
assay when immune-cell staining was considered, the overall concordance was less.
Figure 1 shows the distribution of the NSCLC tumor samples stained when PD-L1 expres-
sion for each sample is determined in relation to the selected threshold for the assay used.
Thirty-eight of the original 40 tumor samples were assessed in this analysis. Although
the 28-8 and 22C3 assays, each with its 1% threshold, determined PD-L1 expression to
be above the threshold for 26 (60.5%) of the 38 tumor samples, these were not the same
26 samples in each instance–each assay determined PD-L1 expression as positive for one
sample that the other assay did not. The SP263 assay determined PD-L1 expression for 20
tumor samples (52.6%) to be above the 25% threshold for that assay. It is no surprise that
the SP263 assay assigned fewer samples into the positive-expression group for this higher
threshold. Although the SP142 assay was shown to stain fewer tumor cells, 30 of 38 (52.6%)
tumor samples were allocated at or above the TC1/IC1 threshold. This is because this assay
allows for a positive expression determination based on immune-cell staining when tumor-
cell staining is below the threshold; fewer samples were allocated over the threshold due to
tumor-cell staining with this assay (Fehrenbacher 2016). These findings are interesting, but
not surprising. They do, however, demonstrate that a given sample may be allocated above
or below a therapeutic threshold, depending on which assay–drug combination would have
been used.
The key factor in considering this difficult question of so-called harmonization is whether
it is possible to use a single staining assay but read that assay according to the different
scoring algorithms associated with a therapeutic decision for different drugs. The outcome
of this variation from trial-validated practice is shown in Table 2. It is clear that using
alternative scoring systems with any particular assay leads to above-threshold allocation
for fewer samples, when compared with the use of a particular assay and its associated
100
22C3
90 28-8
SP142
80 SP263
% Tumor Cell Staining

70
60
50
40
30
20
10
0
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39
Cases
100
22C3
90 28-8
SP142
80
% Immune Cell Staining
SP263
70
60
50
40
30
20
10
0
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39
Cases
Figure 2. Comparability of programmed cell death ligand-1 (PD-L1) staining on tumor and immune cells among four
trial-validated PD-L1 immunohistochemistry assays. Each dot represents the mean score of 3 pathologists. Reprinted with
permission from Hirsch FR et al, J Thorac Oncol. 2017;12(2):208-222.
scoring algorithm. In summary, when each of the SP263, 28-8, and 22C3 assays used the
other two assays cut-off of 1% or 25%, seven instances of difference allocation were seen
(concordance ranged from 81.6% to 94.7%). When samples stained using the SP142 assay
but read using the threshold of 1% associated with the 28-8 and 22C3 assays or the threshold
of 25% associated with the SP263 assay, only 63.2% to 65.8% of samples were concordantly
allocated. When the rules for TC1/IC1 scoring designed for use with the SP142 assay were
applied to sections stained using the 22C3, 28-8, and SP263 assays, only 81.6% to 86.8% of
the sections were concordantly allocated to the same category.
These comparisons indicate that, although there are rough similarities in the performance
of three of the assays (i.e., 28-8, 22C3, and SP263), the SP142 assay behaves differently. This
is not surprising since the latter assay was developed to optimise immune cell as well as
tumor cell staining. The data shown in Figure 2 indicate that the pairing of the drug and the
trial validated threshold should not be broken, otherwise significantly different groups of
patients would be treated. However, if any single assay is chosen, and stained sections are
read in different ways according to the threshold paired to a drug, there is still the potential
for different treatment decisions to be made (approximately 5-19% variance), depending on
which assay was used.
Again, these data are from phase 1 of the BluePrint project, which used a small number of
cases and only three industry pathologists expert in their company’s assays (Hirsch 2017). A
much larger BluePrint 2 study is underway to validate these findings on real-world NSCLC
samples. A wide range of sample types, reflecting typical lung cancer diagnostic practice,
have been collected from several laboratories around the world. The five trial-validated
assays discussed in this Atlas will be used to determine PD-L1 expression, and results will be
determined by practicing thoracic pathologists from five continents. Interobserver variability
will be tested, and staining differences will be compared. While we await these results, the
oncology community must discuss the degree to which variance in PD-L1 expression testing
is acceptable in clinical practice. In reality, complete harmonization cannot be achieved.
Deviating from Trial-Validated Standards: No Good Solutions

One practical solution to the issue of PD-L1 IHC test harmonization would be to use a single
trial-validated, commercially produced assay but use several thresholds and/or scoring algo-
rithms to assess therapeutic options. Several comparison studies suggest that this method
may result in 5% to 15% of patients receiving different treatment. However, the technical
requirements for the staining reagents and procedures are consistent and standardized with
this approach (Table 3).
Table 3. Potential Immunohistochemistry (IHC) Testing Choices for Determination of Programmed Cell
Death Ligand-1 (PD-L1) Expression
Definition of PD-L1 Risk to

Scenario Drug Anti–PD-L1 Assay Expression Positivity Outcome Patient?
#1 By choice Assay validated for Definition validated in Predictable based Known

drug in trial trial for drug upon trial data
#2 By choice Any trial-validated Definition validated in Uncertain Not known

assay trial for drug of choice
#3 By choice Any trial-validated Definition validated Very uncertain Not known

assay with each assay
#4 By choice Laboratory- Unknown Extremely Not known

developed test uncertain
(LDT) using any
antibody clone
Scenario #1 represents laboratory repetition of what was done in the clinical trial. This is the only approach for which we have
clinical outcome data. Scenario #2 is a probable alternative for many laboratories. One of the trial-validated assays is used, but
the IHC stains are read in a way to allow several cut-offs to be assessed. Scenario #3 is not recommended. The cut-off used must
be that associated with the drug/indication or line of therapy. (Figure 1 illustrates the danger of defining a treatment group
according to an inappropriate cut-off for the drug being used.) Scenario #4. LDTs can be developed that match clinical assays,
but they have no validated cut-offs and, therefore, must be read for the intended drug (as in scenario #2). Data have shown the
variability of LDT performance.
There is no such standardization if laboratories continue to deviate from the trial-val-

idated approach to PD-L1 IHC for a given treatment plan. LDTs have no standardization
regardless of whether they use a commercial assay antibody clone or other PD-L1 IHC anti-
body clones. Examples of such LDTs have demonstrated differing staining results (Velcheti
2014; Sheffield 2016; McLaughlin 2016; Neuman 2016). Furthermore, approximately 50%
of the LDTs developed in different laboratories and used in the French harmonization
study showed significant discordance with the data generated using trial-validated assays
(Adam J et al, WCLC 2016). There are generic guidelines for the development of diagnostic
IHC LDTs (Fitzgibbons 2014; Lin 2014), but the technical verification process is very rigor-
ous and might not always be followed precisely. Also, a development standard for LDTs—be
it using a positive control tissue or an existing trial-validated assay—is undetermined. It is
likely that LDTs will continue to be highly variable, and there are no data to confirm that
an LDT has any predictive value if used to select patients for therapy (Table 3) (see Chapters
3 and 9 for details about LDT validation theories). The LDT route would seem to have the
most potential for uncertainty for our patients. Although it is not impossible to develop
an LDT that would match the performance of a trial validated assay (Rimm 2017, Adam J
et al), the development process is demanding and there is no guarantee of consistent success.
Participation in EQA schemes will be essential to ensure that any laboratory, regardless of
how it performs PD-L1 IHC testing, provides a test that is safe and effective.
Conclusion
There are few good data on which to base any firm recommendation for harmonization of
PD-L1 IHC testing. The gold-standard approach, for which there are abundant data—includ-
ing clinical outcomes data—gives the oncologist and the patient an informed prediction of
the likelihood of response and of progression-free and overall survivals for a therapy based
on the assay and corresponding scoring used. Any other approach is less certain, less well
informed, and not clinically validated, although a certain amount of analytical validation
has been attempted. The possibility of assay harmonization is dependent on the amount
of risk oncologists and patients are willing to accept. The use of certain alternative trial-
validated assays (interchanging 28-8, 22C3, and SP263) suggests a possible 5% to 15% loss
of predictive performance. It is surprising that LDTs are being widely used to make clinical
decisions in an era when oncology practice is otherwise so heavily driven by evidence from
clinical trials and especially when considering that PD-L1 IHC as a predictive biomarker has
received considerable criticism. Much more data, including clinical treatment responses,
are required before alternative practices can be determined so that optimal treatment of
patients is not compromised.
12
Implementation of PD-L1 Testing for
Personalized Therapy for Lung Cancer
By Ming S. Tsao, Andrew G. Nicholson, and Fred R. Hirsch
During the past decade, the treatment outlook for patients with advanced-stage lung cancer
has transformed from nihilistic to optimistic. Many patients today receive less toxic thera-
pies and experience longer disease control and survival, and there is a potential for cure
for some patients. These advances partially have been made possible by the development of
personalized therapy based on the molecular characteristics of individual patient’s tumor.
With personalized therapy comes increased and more complex testing. The role of patholo-
gists in routine diagnosis of lung cancer, therefore, has also significantly changed. Optimal
processing of biopsy tissues, stratification of the cut sections, and prioritization of biomarker
analysis are all essential components of both pathologic workflow and diagnosis.
In the clinical practice setting, selection of targeted therapies for patients with non-
small cell lung cancer (NSCLC) requires testing for EGFR mutations and rearrangements
or fusion protein expression involving the ALK and ROS1 genes (Lindeman 2013). EGFR
testing is conducted using DNA isolated from plasma or tumor tissue (Tan 2016). For tissue,
this usually requires a large number (10 or more) of unstained tissue or cytology cell-block
sections. In contrast, current ALK and ROS1 testing is mostly performed by immunohis-
tochemistry (IHC) and/or fluorescence in situ hybridization and requires only one or two
unstained sections (Tsao 2016). Guidelines recommend molecular testing of EGFR, ALK,
and ROS1 aberrations only for patients with adenocarcinoma or non-small cell carcinoma
(NSCC) when an adenocarcinoma component cannot be excluded (e.g., in biopsy samples),
or for patients with squamous cell carcinoma who have high risk of an EGFR, ALK, and/or
ROS1 mutation or rearrangement (e.g., never or light smokers or young women, particularly
with Asian ethnicity) (Lindeman 2013). Although EGFR and ALK genomic aberrations have
been reported in squamous cell carcinoma, routine testing is not recommended based on
low prevalence and cost effectiveness. The previous paradigm of excluding squamous cell
carcinoma for biomarker testing has changed; PD-L1 testing is also required for squamous
cell carcinoma (see Chapter 2 for details).
Strategies for Sample Collection

The addition of new biomarkers for testing has the greatest effect on small-biopsy specimens
due to the limited amount of tissue material per specimen and to the unavoidable loss of
tissue during repeated re-cutting of the paraffin blocks (Kim 2014). A clear guideline on
preservation of biopsy tissue for predictive biomarker testing has been outlined in a collab-
orative effort by the International Association for the Study of Lung Cancer, the American
Thoracic Society, and the European Respiratory Society, as well as by the World Health
Organization (Travis 2011; Travis 2015). It is critical that pathology laboratories develop
strategies for integrating molecular biomarker testing into their routine tissue-processing
workflow and minimize the number of ancillary special stains performed for the diagnosis
and classification of a tumor. The use of two to four stains (thyroid transcription factor-1 and
P40 or P63, with or without mucin and cytokeratin 5) is usually sufficient to classify poorly
differentiated NSCC as either favoring squamous cell carcinoma or adenocarcinoma (Loo
2010; Rekhtman 2011). Additional IHC stains required to diagnose or classify neuroendo-
crine carcinomas (e.g., CD56, chromogranin, and synaptophysin) are usually decided based
on the initial hematoxylin and eosin (H&E) staining appearances of the tumor (Travis 2011;
Thunnissen 2017). Consequently, only two to four additional unstained sections are neces-
sary for routine histologic diagnosis of common lung cancers, which should provide ample
additional sections for further biomarker testing, including PD-L1 expression (Figure 1).
Resection Biopsy
Core/LN Cytology
sampling samples
Diagnostic IHC work-up (TTF-1, mucin, P40/P63, CK5, NE markers)
SCLC, SQCC, ADC, NSCC favor ADC,

LCNEC NSCC favor SQCC
carcinoids ADSQCC, NSCC NOS
NEC+ADC, SQCC
never/light smokers
PD-L1 Testing*
28-8/22C3/SP263/ Tyrosine
73-10/SP142 or LDT EGFR / ALK / ROS1 (Positive)
Kinase
Testing inhibitors
(Negative)
28.8 22C3 SP263 73-10 SP142 PD-L1 Testing*

28-8/22C3/SP263/
NIVO PEMB DURVA AVELU ATEZO 73-10/SP142 or LDT
Figure 1. Immunohistochemistry (IHC) tests that are integral to diagnostic considerations in the treatment of
patients with of lung cancer. LN = lymph node, TTF-1 = thyroid transcription factor-1, CK5 = cytokeratin 5, NE =
neuroendocrine, SCLC = small cell lung cancer, LCNEC = large cell neuroendocrine carcinoma, SQCC = squamous
cell carcinoma, NSCC = non-small cell carcinoma, ADC = adenocarcinoma, ADSQCC = adenosquamous cell car-
cinoma, NEC = neuroendocrine cancer, PD-L1 = programmed cell death ligand-1, TKI = tyrosine kinase inhibitor,
EGFR = epidermal growth factor receptor, ALK = anaplastic lymphoma kinase, NIVO = nivolumab, PEMB = pembro-
lizumab, DURVA = durvalumab, AVELU = avelumumab, ATEZO = atezolizumab, LDT = laboratory developed test.
*Only the 22C3 assay is required as a companion diagnostic for first-line and second/third- line pembrolizumab therapy. The
other assays are for clinical trials or complementary diagnostics.
IMPLEMENTATION OF PD-L1 TESTING FOR PERSONALIZED THERAPY FOR LUNG CANCER 111
Laboratories may prepare 10 or more unstained sections (in addition to the one section
needed for H&E staining) at the time of initial paraffin block cutting, for ancillary studies
including biomarker testing. The alternative approach is for the laboratory to cut additional
sections for ancillary and biomarker tests following a pathologist’s initial assessment of the
H&E slides (Figure 2). Both approaches have their advantages and disadvantages. The first
approach may result in the creation of unnecessary sections for non-neoplastic or nondi-
agnostic samples, and adequate storage space for all additional samples must be available.
The second approach may result in increased turnaround time for the initial diagnosis and
biomarker test results. The biomarker testing strategy—whether reflex testing ordered by the
pathologist or testing ordered by the oncologist (also Chapter 11 for details)—also is relevant
to the sample preparation process. As mentioned previously, it is important to remember
that unstained sections older than 6 weeks might not be usable for most techniques involved
in molecular testing, including IHC, fluorescence in situ hybridization, or DNA sequencing.
A. Biomarker/molecular testing sections prepared together with H&E section

Biopsy: suspicious for
H & E staining for lung cancer
histological diagnosis
For IHC H&E

1
H&E
NSCC
For molecular testing
unstained sections
B. Biomarker/molecular testing sections prepared after initial H&E assessment 2-4

Diagnostic
stains
H & E staining for
histological diagnosis
12 +
Re-sectioning
1-2 ALK IHC

For IHC ± FISH
1-2
ROS1 IHC
± FISH
For molecular
testing EGFR
2
PD-L1
Figure 2. Strategies for maximizing tissue for molecular testing. Unstained sections for ancillary diagnostic immunohistochem-
istry (IHC) of molecular testing may be prepared upfront (A) or after initial histologic assessment of the hematoxylin and eosin
(H&E) stained sections. The number of unstained sections to be prepared is determined by the pathology departmental or
institutional strategy for optimal tissue use and for shortest turnaround times for initial diagnosis and biomarker testing results.
NSCC = non-small cell carcinoma, ALK = anaplastic lymphoma kinase, FISH = fluorescence in situ hybridization, EGFR =
epidermal growth factor receptor, PD-L1 = programmed cell death ligand-1.
Any treatment and biomarker testing algorithm currently proposed (Figure 3) may
become rapidly outdated because of the rapid evolution of the field. However, current need
for PD-L1 testing is strongly influenced by therapeutic decisions. The most recent addi-
tion to the therapies for which PD-L1 testing plays an important role is pembrolizumab,
which is approved as first-line therapy for patients with a PD-L1 Tumor Proportion Score
of 50% or greater and no EGFR mutation or ALK rearrangement. However, determination
of PD-L1 status may not be necessary until after the targeted therapy options are exhausted
for patients with tumors harboring EGFR, ALK, or ROS1 aberrations, which are primarily
treated using their respective tyrosine kinase inhibitors. Pathologists may wish to defer
PD-L1 testing in patients with these aberrations until it is requested by the oncologist.
However, this strategy may potentially compromise the availability of tissue when such
testing is requested because any stored unstained sections may no longer be suitable for
PD-L1 staining, or the tissue block may have been exhausted for EGFR testing.
Nonsquamous Cell Squamous Cell
ROS1+ ALK EGFR PD-L1 ≥ PD-L1 1- PD-L1 PD-L1 ≥ PD-L1 1- PD-L1

Mutation + 50% TPS 49% TPS <1% TPS 50% TPS 49% TPS <1% TPS
Fusion Fusion
First-line
treatment
1st or 2nd Pembroli Pembroli
Chemo ± Bevacizumab Chemo Chemo
Crizotinib generation zumab zumab
EGFR-TKI
Maintenance
(responders only) Bevacizumab
(if eligible)
Pemetrexed
(if eligible)
Second/third-line
treatment
T790M+ T790M–
Chemo Alectinib or Chemo ± Nivolumab or Chemo Nivolumab or

ceritinib Osimertinib Atezolizumab ± Ram Atezolizumab
± Ram Ramucirumab
Pembrolizumab Pembrolizumab
Nivolumab or Nivolumab or
Atezolizumab Atezolizumab
Figure 3. A standard of care treatment algorithm for patients with advanced non-small cell lung cancer proposed in January
2017. ALK = EGFR = epidermal growth factor receptor, PD-L1 = programmed cell death ligand-1, TPS = tumor proportion
score, TKI = tyrosine kinase inhibitor, Chemo = chemotherapy.
For patients with NSCLC who do not have EGFR, ALK, or ROS1 aberrations and who
have received first- or second-line treatment, only pembrolizumab requires determination
of PD-L1 expression (using a 1% threshold) to qualify for treatment. However, depending
on the international region, PD-L1 testing may be used by treating oncologists to inform the
use of nivolumab therapy for patients with non-squamous NSCLC. We will likely witness
more changes to the PD-L1 testing algorithm once results of the ongoing phase II and III
trials of all five anti–PD-1/PD-L1 therapies in various settings become available.
IMPLEMENTATION OF PD-L1 TESTING FOR PERSONALIZED THERAPY FOR LUNG CANCER 113
Conclusion
The current testing algorithm for PD-L1 is evolving and could change dramatically once
results from ongoing trials become available. Nevertheless, pathology laboratories must be
ready to update strategies regarding tissue management and unstained sample preparation
to accommodate PD-L1 testing, as well as strategies on prioritization of testing for the vari-
ous biomarkers encountered in routine clinical practice.
Summary and Future Perspectives
By Yasushi Yatabe, Ming S. Tsao, Keith M. Kerr, Sanja Dacic, and Fred R. Hirsch 13
The emergence of immune checkpoint inhibitors has ushered in dramatic yet exciting prog-
ress in oncology practice, especially for patients with advanced non-small cell lung cancer
(NSCLC). As these agents do not directly target and kill the tumor cells, but instead reacti-
vate a patient’s own immune system to target the cancer cells, toxicity has generally been
mild. Treatment with immune checkpoint inhibitor therapies may result in some unique
clinical features associated with tumor response, adverse effects, and long-term survival.
Characteristic long-term durable response and significant improvement in survival rates in up
to 20% of treated advanced-stage NSCLC patients have been observed. To date, PD-L1 immu-
nohistochemistry (IHC) remains the best validated biomarker for predicting clinical benefit
from anti-PD-1/PD-L1 therapies, as demonstrated in many clinical trials. However, this bio-
marker is different from those for other molecular targeted drugs as summarized in Table 1.
Particularly, some responses to immune checkpoint inhibitors have been observed in patients
who have (or appear to have) low or negative PD-L1 IHC. Most likely this is related, for the
most part, to heterogeneous expression in tumors and biopsy sampling error, and to the fact
that PD-L1 expression is a biologic continuum, such that the creation of ‘positive’ and ‘negative’
Table 1. Differences between PD-L1 IHC and other biomarkers
EGFR/ALK/ROS1 PD-L1 IHC
Result Binary Continuum

Cut-off value /’Positive
Independent of the drugs Different per the drugs
test’
Categories of diagnostic Companion diagnostic Companion and complementary
tests tests diagnostic tests
Distribution in tumor Homogeneous Frequently heterogeneous

Response in biomarker- No responses in vast Responses in a subset, with outcome
negative patients majority of patients similar to the ‘positive’ patients
groups defined by a cut-off does not create two distinct categories that each include patients
who are equally likely or unlikely to benefit from therapy. Thus, there is room to develop
additional biomarkers to either replace PD-L1 IHC or, more likely, enhance the predictive
power of this assay for selecting patients for anti-PD-1/PD-L1 therapy.
Mutation burden in the tumor has been proposed as a predictive biomarker for immune
checkpoint inhibitor therapy. A high non-synonymous mutational load may lead to the
expression of neoantigens, which, if immunogenic, may in turn lead to the development of
tumor-specific T-cell immune responses. Neoantigens are not necessarily immunogenic,
but higher antigen frequency is more likely to lead to a more immunogenic tumor. Despite
this hypothesis, the very existence of tumors with high mutational burden implies that
such tumors have developed a mechanism to escape immune surveillance and progress to
clinical presentation.
The Cancer Genome Atlas project has demonstrated that lung cancers are among those
with the highest mutation rate (Lawrence 2013). Immune-inhibitory checkpoints may be
one such mechanism negating an existing tumor-specific immune response from destroy-
ing immunogenic malignant cell clones. Consequently, tumors with high mutation loads
are expected to respond well to such therapies, accepting the caveat regarding immunoge-
nicity already mentioned. In a non-randomized small cohort study, tumors with a higher
non-synonymous mutation burden showed an improved objective response rate, durable
clinical benefit, and progression-free survival after pembrolizumab treatment (Rizvi
2015). The higher responses to nivolumab observed in smokers could be explained by this
hypothesis, as tumor mutation burden is high in smokers tumors (Hellman 2014; Govindan
2012). Peters et al reported that tumor mutation burden enhanced the predictive power
of PD-L1 IHC for selecting patients who benefit from first-line therapy with nivolumab
(Peters 2017). In contrast, lung cancer patients with EGFR-mutant tumors (known to have
low mutation loads) showed lower response rates than those with wild-type tumors, as
reported in subset analyses of the nivolumab, pembrolizumab, and atezolizumab trials
(Borghaei 2015; Garon 2015b; Rittmeyer 2017). These biologic differences in relation to
tobacco carcinogenesis are also related to the two-compartment model of lung cancer
(Figure 1) (Travis 2015). Central, bronchogenic tumors are mostly related to tobacco carci-
nogenesis; they tend to be squamous or small cell carcinomas and are most often genetically
complex cancers. Tumors arising from the peripheral lung epithelial compartment—the so-
called terminal respiratory unit—may or may not be related to tobacco carcinogens. When
they are not, they tend to be genetically less complex, with a low mutational burden and a
high likelihood of being oncogene-addicted cancers driven by alterations such as EGFR muta-
tion or ALK or ROS1 fusion genes. Therapeutic strategies are essentially different between
the tumors of the two compartments. Genetically complex cancer would potentially be a
good target for immune checkpoint inhibitors. Genetically less complex tumors are less
immunogenic and less responsive to immune checkpoint inhibition, but are generally very
responsive to tyrosine kinase inhibitors targeting their oncogenic drivers.
The above hypothesis suggests that not only should the tumor be immunogenic (muta-
tional burden), but also, the tumor-specific immune response must exist in order that it
may be activated and released from inhibition by the immune checkpoint (PD-1/PD-L1
interaction), the action of an immune checkpoint inhibitor. Thus, some assessment of the
SUMMARY AND FUTURE PERSPECTIVES 117
Genetically
complex cancer
Mutation burden
SCLC, SQC
Central Smoker
airway KRA
compartment Sa
ctiv
atio
n
Other driver mutations

ADC
ion
ivat
Terminal act
FR
respiratory EG
unit
Never Genetically
smoker simple cancer
Figure 1. Genetic complexity and the concept of the two-compartment model in the putative molecular pathogenesis of
lung cancer. Anatomically, lung epithelial cells are divided into two compartments that are associated with lung function.
The central airway compartment functions mainly for air conducting, while respiratory exchange is made in the terminal
respiratory unit of the peripheral compartment. Carcinogens from smoking appear to target both central and peripheral
airways, although the magnitude is weighted more on the central compartment. Long-term smoking causes mutations
across whole genomes, leading to genetically complex tumors with high mutation burden. In contrast, EGFR is mutated by
unidentified factors, which appear to specifically target the terminal respiratory unit. EGFR mutation occurs in the terminal
respiratory unit where smoking has less effect.
immune response in the tumor microenvironment (the inflamed tumor) may also act as a
predictive biomarker (Blank 2016). This approach could involve assessment of actual immune
cell populations in the tumor microenvironment (Hegde 2016; Teng 2015). An alternative
approach has been to examine mRNA expression profiles of immune-related genes in the
tumor as a measure of immunologic activity in the tumor microenvironment (Fehrenbacher
2016; Chen 2016). The consideration of immune cell activity in the tumor microenvironment
mentioned above is also reflected in the way in which PD-L1 IHC has been assessed in some
clinical trials. Recent results of atezolizumab (POPLAR and OAK) trials (Fehrenbacher 2016;
Rittmeyer 2017) showed that PD-L1 expression in tumor-infiltrating immune cells, in the
absence of tumor cell PD-L1 expression, also predicted response to therapy. Tumors showing
≥50% tumor cell PD-L1 staining (TC3) rarely show ≥10% immune cell PD-L1 staining (IC3)
and vice versa. While the atezolizumab-associated SP142 assay demonstrates staining char-
acteristics that may facilitate the scoring of PD-L1 expression on the immune cells (Chapter
6), PD-L1 scoring on immune cells has not been found to have predictive value using other
assays and drugs. The role of immune cell PD-L1 expression levels as a predictive marker
remains worthy of further studies.
To date, the application of the PD-L1 IHC assays has been limited to some immune check-
point monotherapies, mainly in second- or greater-line indications. The use of anti-PD-1/
PD-L1 agents in first-line is now accelerating, driven by PD-L1 IHC biomarker selection as
shown in the KEYNOTE 024 study [Reck 2017]. This is a practice-changing development that
will boost the practice of PD-L1 IHC testing, at least for the foreseeable future. Clinical trials
of these inhibitors are now emerging, using combinations of immune checkpoint inhibi-
tors, or immune checkpoint inhibitors with traditional cytotoxic chemotherapy, where the
role of PD-L1 testing is yet to be defined. It remains to be seen whether PD-L1 IHC will be
replaced by mutational burden or tumor inflammation assessment, or some other biomarker
strategy. Or, perhaps more likely, the predictive power of PD-L1 IHC may be enhanced by
the addition of another test. With intensive efforts to further improve the personalization
of immunotherapies, there is little doubt that in the future, additional and new biomarkers
for immune checkpoint inhibitors will be developed. It remains to be seen whether or not
an increasingly complex, and expensive, biomarker testing strategy will provide significant
improvement over the relatively simple, yet imperfect, PD-L1 IHC that is validated by clini-
cal trial outcomes.
REFERENCES 119
References
Adam J, Rouquette I, Damotte D, et al. PL04a.04: Multi- Blank CU, Haanen JB, Ribas A, Schumacher TN. Cancer
centric French Harmonization Study for PD-L1 IHC Test- Immunology. The “cancer immunogram.” Science. 2016;
ing in NSCLC. J Thorac Oncol. 2017;12(1 suppl):S11-S12. 352(6286):658-660.
Ameratunga M, Asadi K, Lin X, et al. PD-L1 and Tumor Blind C, Koepenik A, Pacyna-Gengelbach M, et al. Anti-
Infiltrating Lymphocytes as Prognostic Markers in genicity testing by immunohistochemistry after tissue
Resected NSCLC. PLoS One. 2016;11(4):e0153954. oxidation. J Clin Pathol. 2008;61(1):79-83.
Ansell SM, Lesokhin AM, Borrello I, et al. PD-1 blockade Bonnas C, Specht K, Spleiss O, et al. Effects of cold
with nivolumab in relapsed or refractory Hodgkin’s lym- ischemia and inflammatory tumor microenviron-
phoma. N Engl J Med. 2015;372(4):311-319. ment on detection of PI3K/AKT and MAPK pathway
activation patterns in clinical cancer samples. Int J
Antonia S, Goldberg SB, Balmanoukian A, et al. Safety
Cancer. 2012;131(7):1621-1632.
and antitumour activity of durvalumab plus tremelim-
umab in non-small cell lung cancer: a multicentre, phase Borghaei H, Paz-Ares L, Horn L, et al. Nivolumab versus
1b study. Lancet Oncol. 2016;17(3):299-308. docetaxel in advanced nonsquamous non-small-cell lung
cancer. N Engl J Med. 2015;373(17):1627-1639.
Antonia SJ, Kim S-W, Spira AI, et al. Safety and clini-
cal activity of durvalumab (MEDI4736), an anti-PD-L1 Boyerinas B, Jochems C, Fantini M, et al. Antibody-
antibody, in treatment-naïve patients with advanced dependent cellular cytotoxicity activity of a novel
non-small-cell lung cancer. J Clin Oncol. 2016;34(15_ anti-PD-L1 antibody avelumab (MSB0010718C) on
suppl):abstr 9029. human tumor cells. Cancer Immunol Res. 2015;3(10):
Antonia SJ, López-Martin JA, Bendell J, et al. Nivolumab 1148-1157.
alone and nivolumab plus ipilimumab in recurrent Brahmer J, Reckamp KL, Baas P, et al. Nivolumab versus
small-cell lung cancer (CheckMate 032): a multicentre, docetaxel in advanced squamous-cell non-small-cell lung
open-label, phase 1/2 trial. Lancet Oncol. 2016;17(7): cancer. N Engl J Med. 2015;373(2):123-135.
883-895.
Brambilla E, Le Teuff G, Marguet S, et al. Prognostic effect
Armand P, Shipp MA, Ribrag V, et al. Programmed of tumor lymphocytic infiltration in resectable non-
death-1 blockade with pembrolizumab in patients with small-cell lung cancer. J Clin Oncol. 2016;34(11):1223-
classical Hodgkin lymphoma after brentuximab vedotin 1230.
failure. J Clin Oncol. 2016 Jun 27. [Epub ahead of print]
Broderick J. Atezolizumab approaches approval for lung
Atkins MB, Lotze MT, Dutcher JP, et al. High-dose recom- cancer. Cure. April 11, 2016. Available at http://www.
binant interleukin 2 therapy for patients with metastatic curetoday.com/articles/atezolizumab-approaches-
melanoma: analysis of 270 patients treated between 1985 approval-for-lung-cancer. Accessed March 10, 2017.
and 1993. J Clin Oncol. 1999;17(7):2105-2116.
Butts C, Murray N, Maksymiuk A, et al. Random-
Azuma K, Ota K, Kawahara A, et al. Association of PD-L1 ized phase IIB trial of BLP25 liposome vaccine in
overexpression with activating EGFR mutations in stage IIIB and IV non-small-cell lung cancer. J Clin
surgically resected nonsmall-cell lung cancer. Ann Oncol. 2005;23(27):6674-6681.
Oncol. 2014;25(10):1935-1940.
Butts C, Socinski MA, Mitchell PL, et al. Tec-
Baas P, Garon EB, Herbst RS, et al. Relationship between emotide (L-BLP25) versus placebo after chemoradio-
level of PD-L1 expression and outcomes in the KEY- therapy for stage III non-small-cell lung cancer (START):
NOTE-010 study of pembrolizumab vs docetaxel for
a randomised, double-blind, phase 3 trial. Lancet Oncol.
previously treated, PD-L1–Positive NSCLC. J Clin Oncol.
2014;15(1):59-68.
2016;34(15_suppl): abstr 9015.
Camidge DR, Pao W, Sequist LV. Acquired resistance to
Bakker W, Nijhuis-Heddes JM, van der Velde EA. Post-
TKIs in solid tumours: learning from lung cancer.
operative intrapleural BCG in lung cancer: a 5-year fol-
Nat Rev Clin Oncol. 2014;11(8):473-481.
low-up report. Cancer Immunol Immunother. 1986;22(2):
155-159. Chen DS, Irving BA, Hodi FS. Molecular pathways: next-
generation immunotherapy--inhibiting programmed
Besse B, Johnson ML, Janne PA, et al. Phase II, single-
death-ligand 1 and programmed death-1. Clin Cancer
arm trial (BIRCH) of atezolizumab as first-line or sub-
Res. 2012;18(24):6580-6587.
sequent therapy for locally advanced or metastatic
PD-L1–selected non-small cell lung cancer (NSCLC). Chen PL, Roh W, Reuben A, et al. Analysis of immune
Eur J Cancer. 2015;51(suppl 3):S717–S718. signatures in longitudinal tumor samples yields insight
into biomarkers of response and mechanisms of resis- Dolled-Filhart M, Roach C, Toland G, et al. Development
tance to immune checkpoint blockade. Cancer Discov. of a Companion Diagnostic for Pembrolizumab in Non-
2016;6(8):827-837. Small Cell Lung Cancer Using Immunohistochemistry
for Programmed Death Ligand-1. Arch Pathol Lab Med.
Cheung CC, Garratt J, Won J, et al. Developing ALK
2016 Aug 23 [Epub ahead of print].
Immunohistochemistry and In Situ Hybridization
Proficiency Testing for Non-Small Cell Lung Cancer Dong H, Strome SE, Salomao DR, et al. Tumor-associated
in Canada: Canadian Immunohistochemistry Quality B7-H1 promotes T-cell apoptosis: a potential mechanism
Control Challenges and Successes. Appl Immunohistochem of immune evasion. Nat Med. 2002;8(8):793-800.
Mol Morphol. 2015;23(10):677-681.
Fehrenbacher L, Spira A, Ballinger M, et al. Atezolizumab
Cogswell J, Inzunza HD, Wu Q, et al. An Analytical versus docetaxel for patients with previously treated
Comparison of Dako 28-8 PharmDx Assay and an E1L3N non-small-cell lung cancer (POPLAR): a multicenter,
Laboratory-Developed Test in the Immunohistochemical open-label, phase 2 randomised controlled trial. Lancet.
Detection of Programmed Death-Ligand 1. Mol Diagn 2016;387(10030):1837-1846.
Ther. 2017;21(1):85-93.
Fischer AH, Schwartz MR, Moriarty AT, et al. Immu-
Coiffier B, Lepage E, Briere J, et al. CHOP chemotherapy nohistochemistry practices of cytopathology laborato-
plus rituximab compared with CHOP alone in elderly ries: a survey of participants in the College of American
patients with diffuse large-B-cell lymphoma. N Engl J Pathologists Nongynecologic Cytopathology Education
Med. 2002;346(4):235-242. Program. Arch Pathol Lab Med. 2014;138(9):1167-1172.
Cooper W, Russell PA, Huot-Marhand P, et al. Intra- and Fitzgibbons PL, Bradley LA, Fatheree LA, et al. Principles
interobserver reproducibility study of PD-L1 biomarker of analytic validation of immunohistochemical assays:
in NSCLC. The Dream study. J Thorac Oncol. 2016;12(1 guideline from the College of American Pathologists
suppl):S814-S815. Pathology and Laboratory Quality Center. Arch Pathol
Cooper WA, Tran T, Vilain RE, et al. PD-L1 expression is Lab Med. 2014;138(11):1432-1443.
a favorable prognostic factor in early stage non-small cell Fridman WH, Pages F, Sautes-Fridman C, Galon J. The
carcinoma. Lung Cancer 2015;89(2):181-188. immune contexture in human tumours: impact on clini-
Cree IA, Booton R, Cane P, et al. PD-L1 testing for lung cal outcome. Nat Rev Cancer. 2012;12(4):298-306.
cancer in the UK: recognizing the challenges for imple- Fyfe G, Fisher RI, Rosenberg SA, Sznol M, Parkinson
mentation. Histopathology. 2016;69(2):177-186. DR, Louie AC. Results of treatment of 255 patients with
Cuppens K, Vansteenkiste J. Vaccination therapy for metastatic renal cell carcinoma who received high-
non-small-cell lung cancer. Curr Opin Oncol. 2014;26(2): dose recombinant interleukin-2 therapy. J Clin Oncol.
165-170. 1995;13(3):688-696.
Curran MA, Montalvo W, Yagita H, Allison JP. PD-1 Garassino M, Vansteenkiste JF, Kim J, et al. Durvalumab
and CTLA-4 combination blockade expands infiltrat- in ≥3rd-line locally advanced or metastatic, EGFR/ALK
ing T cells and reduces regulatory T and myeloid cells wild-type NSCLC: results from the phase 2 ATLANTIC
within B16 melanoma tumors. Proc Natl Acad Sci U S A. study. Presented at World Conference on Lung Cancer;
2010;107(9):4275-4280. December 7, 2016; Vienna, Austria.
Cutz JC, Craddock KJ, Torlakovic E, et al. Canadian ana- Garassino M, Vansteenkiste J, Kim JH, et al. PL04a.03:
plastic lymphoma kinase study: a model for multicenter Durvalumab in ≥3rd-Line Locally Advanced or Meta-
standardization and optimization of ALK testing in lung static, EGFR/ALK Wild-Type NSCLC: Results from
cancer. J Thorac Oncol. 2014;9(9):1255-1263. the Phase 2 ATLANTIC Study. J Thorac Oncol. 2017;12
(1 suppl):S10-S11.
Dako North America. PD-L1 IHC 22C3 pharmDx inter-
pretation manual. http://www.dako.com/us/29109_pd- Garcia-Lora A, Martinez M, Algarra I, Gaforio JJ, Garrido
l1-ihc-22c3-interpretation-manual.pdf. (accessed August F. MHC class I-deficient metastatic tumor variants
13, 2016.). immunoselected by T lymphocytes originate from the
coordinated downregulation of APM components.
Decoster L, Wauters I, Vansteenkiste JF. Vaccination Int J Cancer. 2003;106(4):521-527.
therapy for non-small-cell lung cancer: review of agents
in phase III development. Ann Oncol. 2012;23(6):1387- Garon EB. Current Perspectives in Immunotherapy
1393. for Non-Small Cell Lung Cancer. Semin Oncol. 2015;42
(suppl 2):S11-S18. (a)
Dolled-Filhart M, Locke D, Murphy T, et al. Develop-
ment of a Prototype Immunohistochemistry Assay to Garon EB, Rizvi NA, Hui R, et al. KEYNOTE-001 Inves-
Measure Programmed Death Ligand-1 Expression in tigators. Pembrolizumab for the treatment of non-small-
Tumor Tissue. Arch Pathol Lab Med. 2016;140(11): cell lung cancer. N Engl J Med. 2015;372(21):2018-
1259-1266. 2028. (b)
REFERENCES 121
Garrido F, Cabrera T, Lopez-Nevot MA, Ruiz-Cabello F. Hegde PS, Karanikas V, Evers S. The where, the when,
HLA class I antigens in human tumors. Adv Cancer Res. and the how of immune monitoring for cancer immuno-
1995;67:155-195. therapies in the era of checkpoint inhibition. Clin Cancer
Res. 2016;22:1865-1874.
Gaule P, Smithy JW, Toki M, et al. A quantitative
comparison of antibodies to programmed cell death 1 Hellman MD, Creelan BC, Woo K, et al. Smoking history
ligand 1. JAMA Oncol. 2016 Aug 18. [Epub ahead of print] and response to nivolumab in patients with advanced
NSCLCs. Ann Oncol. 2014;25:iv426–iv470 (suppl 4).
Gettinger S, Rizvi NA, Chow LQ, et al. Nivolumab
monotherapy for first-line treatment of advanced non- Hellman MD, Gettinger SN, Goldman JW, et al. Check-
small-cell lung cancer. J Clin Oncol. 2016;34(25):2980- Mate 012: Safety and efficacy of first-line (1L) nivolumab
2987. (nivo; N) and ipilimumab (ipi; I) in advanced (adv)
NSCLC. J Clin Oncol. 2016;34(suppl):abstr 3001.
Gettinger SN, Horn L, Gandhi L, et al. Overall survival
and long-term safety of nivolumab (anti-programmed Herbst RS, Baas P, Kim DW, et al. Pembrolizumab
death 1 antibody, BMS-936558, ONO-4538) in patients versus docetaxel for previously treated, PD-L1-pos-
with previously treated advanced non-small-cell lung itive, advanced non-small-cell lung cancer (KEY-
cancer. J Clin Oncol. 2015;33(18):2004-2012. NOTE-010): a randomised controlled trial. Lancet.
Giaccone G, Bazhenova LA, Nemunaitis J, et al. A phase 2016;387(10027):1540-1550.
III study of belagenpumatucel-L, an allogeneic tumour Herbst RS, Soria JC, Kowanetz M, et al. Predictive corre-
cell vaccine, as maintenance therapy for non-small cell lates of response to the anti-PD-L1 antibody MPDL3280A
lung cancer. Eur J Cancer. 2015;51(16):2321-2329. in cancer patients. Nature. 2014;515(7528):563-567.
Gibbons DL, Chow LQ, Kim DW, et al. 57O Efficacy, Hirsch FR, McElhinny A, Stanforth D, et al. PD-L1
safety and tolerability of MEDI4736 (durvalumab [D]), Immunohistochemistry assays for lung cancer: Results
a human IgG1 anti-programmed cell death-ligand-1 from phase 1 of the Blueprint PD-L1 IHC assay
(PD-L1) antibody, combined with gefitinib (G): A phase I comparison project. J Thorac Oncol. 2017; 12(2):208-222.
expansion in TKI-naive patients (pts) with EGFR mutant
NSCLC. J Thorac Oncol. 2016;11(4 suppl):S79. Hirsch FR, Suda K, Wiens J, Bunn PA Jr. New and
emerging targeted treatments in advanced non-small-
Govindan R, Ding L, Griffith M, et al. Genomic landscape cell lung cancer. Lancet. 2016;388(10048):1012-1024.
of non-small cell lung cancer in smokers and never-smok-
ers. Cell. 2012;150(6):1121-1134. Hodi FS, O’Day SJ, McDermott DF, et al. Improved
survival with ipilimumab in patients with metastatic
Gown AM. Diagnostic Immunohistochemistry: What melanoma. N Engl J Med. 2010;363(8):711-723.
Can Go Wrong and How to Prevent It. Arch Pathol Lab
Med. 2016;140(9):893-898. Homet Moreno B, Ribas A. Anti-programmed cell death
protein-1/ligand-1 therapy in different cancers. Br J
Gralla RJ, Coon C, Taylor F, et al. Evaluation of disease- Cancer. 2015;112(9):1421-1427.
related symptoms in patients with advanced squamous
non-small cell lung cancer treated with nivolumab or Horn L, Spigel DR, Gettinger SN, et al. Clinical activ-
docetaxel. Presented at World Conference on Lung ity, safety and predictive biomarkers of the engineered
Cancer; September 7, 2015; Denver, CO. antibody MPDL3280A (anti-PDL1) in non-small cell lung
cancer (NSCLC): update from a phase Ia study. J Clin
Gubens MA, Sequist LV, Stevenson J, et al. Phase I/II study Oncol 2015;33(15_suppl):abstr 8029.
of pembrolizumab (pembro) plus ipilimumab (ipi) as sec-
ond-line therapy for NSCLC: KEYNOTE-021 cohorts D Hui R, Gandhi L, Costa EC, et al. Long-term OS for
and H. J Clin Oncol. 2016;34 (suppl; abstr 9027). patients with advanced NSCLC enrolled in the KEY-
NOTE-001 study of pembrolizumab (pembro). J Clin
Gulley JL, Rajan A, Spigel DR, et al. Avelumab for patients Oncol. 2016;34 (15_suppl): abstr 9026.
with previously treated metastatic or recurrent non-
small-cell lung cancer (JAVELIN Solid Tumor): Dose- Hui R, Garon EB, Goldman JW, et al. Pembrolizumab
expansion cohort of a multicentre, open-label, phase 1b as first-line therapy for patients with PD-L1-positive
trial. Lancet Oncol. 2017. pii: S1470-2045(17)30240-1. advanced non-small cell lung cancer: a phase 1 trial. Ann
[Epub ahead of print] Oncol. 2017 Feb 7. [Epub ahead of print]
Hagenbaugh A, Sharma S, Dubinett SM, et al. Altered Hussein M, McCleod M, Chandler J, et al. Safety and
immune responses in interleukin 10 transgenic mice. efficacy of nivolumab in an ongoing trial of a PD-L1+/-
J Exp Med. 1997;185(12):2101-2110. patient population with metastatic non-small cell lung
cancer. Presented at World Conference on Lung Cancer;
Hecht M, Büttner-Herold M, Erlenbach-Wünsch K,
September 7, 2015; Denver, CO.
et al. PD-L1 is upregulated by radiochemotherapy in
rectal adenocarcinoma patients and associated with a Huynh TG, Morales-Oyarvide V, Campo MJ, et al.
favourable prognosis. Eur J Cancer. 2016;65:52-60. Programmed Cell Death Ligand 1 Expression in Resected
Lung Adenocarcinomas: Association with Immune Kerr KM, Tsao MS, Nicholson AG, et al. IASLC Pathol-
Microenvironment. J Thorac Oncol. 2016;11(11): ogy Committee. Programmed death-ligand 1 immuno-
1869-1878. histochemistry in lung cancer: in what state is this art? J
Thorac Oncol. 2015;10(7):985-989.
Ibrahim M, Parry S, Wilkinson D, et al. ALK Immu-
nohistochemistry in NSCLC: Discordant Staining Kikuchi E, Yamazaki K, Torigoe T, et al. HLA class I anti-
Can Impact Patient Treatment Regimen. J Thorac gen expression is associated with a favorable prognosis
Oncol. 2016;11(12):2241-2247. in early stage non-small cell lung cancer. Cancer Sci.
2007;98(9):1424-1430.
Igarashi T, Teramoto K, Ishida M, Hanaoka J, Daigo Y.
Scoring of PD-L1 expression intensity on pulmonary Kim L, Tsao MS. Tumour tissue sampling for lung cancer
adenocarcinomas and the correlations with clinico- management in the era of personalised therapy: what is
pathological factors. ESMO Open. 2016;1(4):e000083. good enough for molecular testing? Eur Respir J. 2014;
44(4):1011-1022.
Ilie M, Falk AT, Butori C, et al. PD-L1 expression in basa-
loid squamous cell lung carcinoma: Relationship to PD-1+ Kim MY, Koh J, Kim S, et al. Clinicopathological analysis
and CD8+ tumor-infiltrating T cells and outcome. Mod of PD-L1 and PD-L2 expression in pulmonary squamous
Pathol. 2016; 29(12):1552-1564. cell carcinoma: Comparison with tumor-infiltrating T
cells and the status of oncogenic drivers. Lung Cancer.
Ilie M, Long-Mira E, Bence C, et al. Comparative study of 2015;88(1):24-33.
the PD-L1 status between surgically resected specimens
and matched biopsies of NSCLC patients reveal major Kirkwood JM, Strawderman MH, Ernstoff MS, Smith
discordances: a potential issue for anti-PD-L1 therapeutic TJ, Borden EC, Blum RH. Interferon alfa-2b adjuvant
strategies. Ann Oncol. 2016;27(1):147-153. therapy of high-risk resected cutaneous melanoma: the
Eastern Cooperative Oncology Group Trial EST 1684. J
Inaguma S, Wang Z, Lasota J, et al. Comprehensive Clin Oncol. 1996;14(1):7-17.
Immunohistochemical Study of Programmed Cell Death
Ligand 1 (PD-L1): Analysis in 5536 Cases Revealed Kitazono S, Fujiwara Y, Tsuta K, et al. Reliability of small
Consistent Expression in Trophoblastic Tumors. Am J biopsy samples compared with resected specimens
Surg Pathol. 2016;40(8):1133-1142. for the determination of programmed death-ligand 1
expression in non-small-cell lung cancer. Clin Lung
Jansen RL, Slingerland R, Goey SH, Franks CR, Bolhuis Cancer. 2015;16(5):385-390.
RL, Stoter G. Interleukin-2 and interferon-alpha in the
Koh J, Go H, Keam B, et al. Clinicopathologic analysis of
treatment of patients with advanced non-small-cell lung
programmed cell death-1 and programmed cell death-
cancer. J Immunother (1991). 1992;12(1):70-73.
ligand 1 and 2 expressions in pulmonary adenocarci-
Jerusalem G, Chen F, Spigel D, et al. JAVELIN Solid noma: comparison with histology and driver oncogenic
Tumor: Safety and Clinical Activity of Avelumab alteration status. Mod Pathol. 2015;28(9):1154-1166.
(Anti-PD-L1) as First-Line Treatment in Patients with
Langer CJ, Gadgeel SM, Borghaei H, et al. Carboplatin
Advanced NSCLC. J Thorac Oncol. 2017; 12:S252.
and pemetrexed with or without pembrolizumab for
Kantoff PW, Higano CS, Shore ND, et al. Sipuleucel-T advanced, non-squamous non-small-cell lung cancer:
immunotherapy for castration-resistant prostate cancer. a randomised, phase 2 cohort of the open-label
N Engl J Med. 2010;363(5):411-422. KEYNOTE-021 study. Lancet Oncol. 2016;17(11):
1497-1508.
Kaufman HL, Russell J, Hamid O, et al. Avelumab in
patients with chemotherapy-refractory metastatic Larkin J, Chiarion-Sileni V, Gonzalez R, et al. Combined
Merkel cell carcinoma: a multicentre, single-group, nivolumab and ipilimumab or monotherapy in untreated
open-label, phase 2 trial. Lancet Oncol. 2016 Oct;17(10): melanoma. N Engl J Med. 2015;373(1):23-34.
1374-1385.
Latchman Y, Wood CR, Chernova T, et al. PD-L2 is a
Kerr KM, Bubendorf L, Edelman MJ, et al. Panel second ligand for PD-1 and inhibits T cell activation. Nat
Members. Second ESMO consensus conference on Immunol. 2001;2(3):261-268.
lung cancer: pathology and molecular biomarkers for Lawrence MS, Stojanov P, Polak P, et al. Mutational
non-small-cell lung cancer. Ann Oncol. 2014;25(9): heterogeneity in cancer and the search for new cancer-
1681-1690. associated genes. Nature. 2013;499(7457):214-218.
Kerr KM, Hirsch FR. Programmed Death Ligand-1 Immu- Le DT, Uram JN, Wang H, et al. PD-1 Blockade in
nohistochemistry: Friend or Foe? Arch Pathol Lab Med. tumors with mismatch-repair deficiency. N Engl J Med.
2016;140:326-331. (a) 2015;372(26):2509-2520.
Kerr KM, Nicolson MC. Non-small cell lung cancer, Li X, Deavers MT, Guo M, et al. The effect of prolonged cold
PD-L1, and the pathologist. Arch Pathol Lab Med. ischemia time on estrogen receptor immunohistochemistry
2016;140:249-254. (b) in breast cancer. Mod Pathol. 2013;26(1):71-78.
REFERENCES 123
Li Y, Xu KP, Jiang D, Zhao J, Ge JF, Zheng SY. Relation- McLaughlin J, Han G, Schalper KA, et al. Quantitative
ship of Fas, FasL, p53 and bcl-2 expression in human Assessment of the Heterogeneity of PD-L1 Expres-
non-small cell lung carcinomas. Int J Clin Exp Pathol. sion in Non-Small-Cell Lung Cancer. JAMA Oncol. 2015;
2015;8(11):13978-13986. 2(1):46-54.
Lin F, Chen Z. Standardization of diagnostic immunohis- Merck KGaA. Merck KGaA discontinues clinical develop-
tochemistry: literature review and geisinger experience. ment program of tecemotide as a monotherapy in stage
Arch Pathol Lab Med. 2014;138:1564-1577. III non-small cell lung cancer. Available at http://www.
prnewswire.com/news-releases/merck-kgaa-discon-
Lindeman NI, Cagle PT, Beasley MB, et al. Molecular
tinues-clinical-development-program-of-tecemotide-
testing guideline for selection of lung cancer patients
as-a-monotherapy-in-stage-iii-non-small-cell-lung-
for EGFR and ALK tyrosine kinase inhibitors: guideline
cancer-274891191.html. Published September 12, 2014.
from the College of American Pathologists, International
Accessed March 8, 2017.
Association for the Study of Lung Cancer, and Association
for Molecular Pathology. J Thorac Oncol. 2013;8(7): Meric-Bernstam F, Akcakanat A, Chen H, et al. Influence
823-859. of biospecimen variables on proteomic biomarkers in
breast cancer. Clin Cancer Res. 2014;20(14):3870-3883.
Loo PS, Thomas SC, Nicolson MC, Fyfe MN, Kerr KM.
Subtyping of undifferentiated non-small cell carcino- Midha A, Sharpe A, Scott M. PD-L1 expression in advanced
mas in bronchial biopsy specimens. J Thorac Oncol. NSCLC: Primary lesions versus metastatic sites and impact of
2010;5(4):442-447. sample age. J Clin Oncol. 2016;34:(15_suppl): abstr 3025.
Lotze MT, Matory YL, Ettinghausen SE, et al. In vivo Miller JW, Roscoe P, Pearce SJ, Ludgate S, Horne NW.
administration of purified human interleukin 2. II. Half Five-year results of a controlled study of BCG immuno-
life, immunologic effects, and expansion of peripheral therapy after surgical resection in bronchogenic carci-
lymphoid cells in vivo with recombinant IL 2. J Immunol. noma. Thorax. 1982;37(1):57-60.
1985;135(4):2865-2875.
Milne CP, Bryan C, Garafalo S, McKiernan M. Comple-
Lynch TJ, Bondarenko I, Luft A, et al. Ipilimumab in com- mentary versus companion diagnostics: apples and
bination with paclitaxel and carboplatin as first-line treat- oranges? Biomark Med. 2015;9(1):25-34.
ment in stage IIIB/IV non-small-cell lung cancer: results
Mok T, Wu YL, Sadowski S, et al. Pembrolizumab (MK-
from a randomized, double-blind, multicenter phase II
3475) versus platinum-based chemotherapy for PD-L1+
study. J Clin Oncol. 2012;30(17):2046-2054.
NSCLC in a phase 3, randomized, open-label study: KEY-
Mahoney KM, Sun H, Liao X, et al. PD-L1 Antibodies NOTE-042. J Thorac Oncol. 2016;11(4 suppl):S142.
to Its Cytoplasmic Domain Most Clearly Delineate Cell
Mok TS, Wu YL, Thongprasert S, et al. Gefitinib or carbo-
Membranes in Immunohistochemical Staining of Tumor
platin-paclitaxel in pulmonary adenocarcinoma. N Engl
Cells. Cancer Immunol Res. 2015;3(12):1308-1315.
J Med. 2009;361(10):947-957.
Maio M, Grob JJ, Aamdal S, et al. Five-year survival rates
Morgan DA, Ruscetti FW, Gallo R. Selective in vitro
for treatment-naive patients with advanced melanoma
growth of T lymphocytes from normal human bone mar-
who received ipilimumab plus dacarbazine in a phase III
rows. Science. 1976;193(4257):1007-1008.
trial. J Clin Oncol. 2015;33(10):1191-1196.
Morgan RA, Dudley ME, Wunderlich JR, et al. Cancer
Marzec M, Zhang Q, Goradia A, et al. Oncogenic kinase
regression in patients after transfer of genetically engi-
NPM/ALK induces through STAT3 expression of immu-
neered lymphocytes. Science. 2006;314(5796):126-129.
nosuppressive protein CD274 (PD-L1, B7-H1). ProcNatl
Acad Sci U S A. 2008;105(52):20852-20857. Motzer RJ, Bacik J, Murphy BA, Russo P, Mazumdar M.
Interferon-alfa as a comparative treatment for clinical
Massard C, Gordon MS, Sharma S, et al. Safety and effi-
trials of new therapies against advanced renal cell carci-
cacy of durvalumab (MEDI4736), an anti-programmed
noma. J Clin Oncol. 2002;20(1):289-296.
cell death ligand-1 immune checkpoint inhibitor, in
patients with advanced urothelial bladder cancer. J Clin Motzer RJ, Escudier B, McDermott DF, et al. Check-
Oncol. 2016;34(26):3119-3125. Mate 025 Investigators. Nivolumab versus everoli-
mus in advanced renal-cell carcinoma. N Engl J Med.
Matthay RA, Mahler DA, Beck GJ, et al. Intratumoral
2015;373(19):1803-1813.
Bacillus Calmette-Guérin immunotherapy prior to
surgery for carcinoma of the lung: results of a prospec- Muro K, Chung HC, Shankaran V, et al. Pembrolizumab
tive randomized trial. Cancer Res. 1986;46(11): for patients with PD-L1-positive advanced gastric cancer
5963-5968. (KEYNOTE-012): a multicentre, open-label, phase 1b
trial. Lancet Oncol. 2016;17(6):717-726.
Maude SL, Frey N, Shaw PA, et al. Chimeric antigen
receptor T cells for sustained remissions in leukemia. National Cancer Institute Surveillance, Epidemiol-
N Engl J Med. 2014;371(16):1507-1517. ogy, and End Results Program. SEER Cancer Statistics
Review, 1975-2011. https://seer.cancer.gov/archive/ associated immune cells density supports distinct intra-
csr/1975_2011/. Updated December 17, 2014. Accessed tumoral microenvironment groups in non-small cell
March 8, 2017. lung carcinoma patients. Clin Cancer Res. 2016;22(24):
6278-6289.
National Comprehensive Cancer Network. Non-small
cell lung cancer. Version 5.2017. https://www.nccn.org/ Patel SP, Kurzrock R. PD-L1 expression as a predic-
professionals/physician_gls/pdf/nscl.pdf. Accessed tive biomarker in cancer immunotherapy. Mol Cancer
March 21, 2017. Ther. 2015;14(4):847-856.
Nemunaitis J, Dillman RO, Schwarzenberger PO, et al. Paulsen EE, Kilvaer TK, Khanehkenari MR, et al.
Phase II study of belagenpumatucel-L, a transforming Assessing PDL-1 and PD-1 in non-small cell lung
growth factor beta-2 antisense gene-modified allogeneic cancer: A novel immunoscore approach. Clin Lung
tumor cell vaccine in non-small-cell lung cancer. J Clin Cancer. 2017;18(2):220-233.
Oncol. 2006;24(29):4721-4730.
Peters S. Impact of tumor mutation burden on the effi-
Neninger Vinageras E, de la Torre A, et al. Phase II ran- cacy of first-line nivolumab in stage IV or recurrent
domized controlled trial of an epidermal growth factor non-small cell lung cancer: An exploratory analysis of
vaccine in advanced non-small-cell lung cancer. J Clin CheckMate 026. Presented at AACR Annual Meeting,
Oncol. 2008;26(9):1452-1458. April 3, 2017; Washington, DC.
Neuman T, London M, Kania-Almog J, et al. A Harmoni- Petersen RP, Campa MJ, Sperlazza J, et al. Tumor infil-
zation Study for the Use of 22C3 PD-L1 Immunohisto- trating Foxp3+ regulatory T-cells are associated with
chemical Staining on Ventana’s Platform. J Thorac Oncol.
recurrence in pathologic stage I NSCLC patients. Cancer.
2016;11:1863-1868.
2006;107(12):2866-2872.
Neumeister VM, Anagnostou V, Siddiqui S, et al. Quan-
Pfizer, Inc. Avelumab Fact Sheet. https://www.
titative assessment of effect of preanalytic cold ischemic
pfizer.com/files/news/asco/Merck-PfizerAlliance_
time on protein expression in breast cancer tissues. J Natl
AvelumabFactSheet_19May2015US.pdf. Published May
Cancer Inst. 2012;104(23):1815-1824.
19, 2015. Accessed March 17, 2017.
Nghiem PT, Bhatia S, Lipson EJ, et al. PD-1 Blockade with
Phillips T, Simmons P, Inzunza HD, et al. Development of
pembrolizumab in advanced Merkel-cell carcinoma. N
an automated PD-L1 immunohistochemistry (IHC) assay
Engl J Med. 2016;374(26):2542-2552.
for non-small cell lung cancer. Appl Immunohistochem
Novotny JF Jr, Cogswell J, Inzunza H, Harbison C, Horak Mol Morphol. 2015;23(8):541-549.
C, Averbuch S. Establishing a complementary diagnostic
for anti-PD-1 immune checkpoint inhibitor therapy. Ann Planchard D, Kim TM, Mazieres J, et al. Dabrafenib in
Oncol. 2016;27:1966-1969. patients with BRAF(V600E)-positive advanced non-
small-cell lung cancer: a single-arm, multicentre, open-
NSCLC Meta-Analyses Collaborative Group. Chemo- label, phase 2 trial. Lancet Oncol. 2016;17(5):642-650.
therapy in addition to supportive care improves sur-
vival in advanced non-small-cell lung cancer: a sys- Portier BP, Wang Z, Downs-Kelly E, et al. Delay to for-
tematic review and meta-analysis of individual patient malin fixation ’cold ischemia time’: effect on ERBB2
data from 16 randomized controlled trials. J Clin Oncol. detection by in-situ hybridization and immunohisto-
2008;26(28):4617-4625. chemistry. Mod Pathol. 2013;26(1):1-9.
Ott PA, Fernandez MEE, Hiret S, et al. Pembrolizumab Powles T, Eder JP, Fine GD, et al. MPDL3280A (anti-
(MK-3475) in patients (pts) with extensive-stage small PD-L1) treatment leads to clinical activity in metastatic
cell lung cancer (SCLC): Preliminary safety and efficacy bladder cancer. Nature. 2014;515(7528):558-562.
results from KEYNOTE-028. J Clin Oncol. 2015;33(15_
Prinsen CF, Klaassen CH, Thunnissen FB. Microarray as
suppl):abstr 7502.
a model for quantitative visualization chemistry. Appl
Ott PA, Hodi FS, Robert C. CTLA-4 and PD-1/PD-L1 Immunohistochem Mol Morphol. 2003;11(2):168-173.
blockade: new immunotherapeutic modalities with dura-
Quoix E, Lena H, Losonczy G, et al. TG4010 immu-
ble clinical benefit in melanoma patients. Clin Cancer Res.
notherapy and first-line chemotherapy for advanced
2013;19(19):5300-5309.
non-small-cell lung cancer (TIME): results from the
Pao W, Iafrate AJ, Su Z. Genetically informed lung cancer phase 2b part of a randomised, double-blind, placebo-
medicine. J Pathol. 2011;223(2):230-240. controlled, phase 2b/3 trial. Lancet Oncol. 2016;17(2):
Pardoll DM. The blockade of immune checkpoints in 212-223.
cancer immunotherapy. Nat Rev Cancer. 2012;12(4): Quon C, Xia X, Smith M, et al. VENTANA anti-PD-L1
252-264. (SP263) rabbit monoclonal antibody: A high specificity
Parra ER, Behrens C, Rodriguez-Canales J, et al. Image and sensitivity anti-human PD-L1 antibody. J Clin Oncol.
analysis-based assessment of PD-L1 and tumor- 2016;34(15_suppl):abstr e14509.
REFERENCES 125
Ramnath N, Tan D, Li Q, et al. Is downregulation of MHC Rizvi NA, Brahmer JR, Ou SH, et al. Safety and clinical
class I antigen expression in human non-small cell lung activity of MEDI4736, an anti-programmed cell death-
cancer associated with prolonged survival? Cancer Immu- ligand 1 (PD-L1) antibody, in patients with non-small cell
nol Immunother. 2006;55(8):891-899. lung cancer (NSCLC). J Clin Oncol. 2015;33 (15_suppl):
abstr 8032.
Ratcliffe MJ, Sharpe A, Midha A, et al. Agreement
between Programmed Cell Death Ligand-1 Diagnostic Rizvi NA, Hellmann MD, Brahmer JR, et al. Nivolumab in
Assays across Multiple Protein Expression Cut-Offs in combination with platinum-based doublet chemotherapy
Non-Small Cell Lung Cancer. Clin Cancer Res. 2017 Jan for first-line treatment of advanced non-small-cell lung
10 [Epub ahead of print]. cancer. J Clin Oncol. 2016;34(25):2969-2979.
Rebelatto MC, Midha A, Mistry A, et al. Development of Rizvi NA, Hellmann MD, Snyder A, et al. Cancer immu-
a programmed cell death ligand-1 immunohistochemi- nology. Mutational landscape determines sensitivity to
cal assay validated for analysis of non-small cell lung PD-1 blockade in non-small cell lung cancer. Science.
cancer and head and neck squamous cell carcinoma. 2015;348(6230):124-128.
Diagn Pathol. 2016;11(1):95.
Rizvi NA, Mazières J, Planchard D, et al. Activity and
Reck M, Coon C, Taylor F, Derosa M, et al. Evaluation of safety of nivolumab, an anti-PD-1 immune checkpoint
overall health status in patients with advanced squamous inhibitor, for patients with advanced, refractory squamous
non-small cell lung cancer treated with nivolumab or non-small-cell lung cancer (CheckMate 063): a phase 2,
docetaxel in CheckMate 017. Ann Oncol. 2015;26(suppl single-arm trial. Lancet Oncol. 2015;16(3):257-265.
9):125-147.
Robert C, Schachter J, Long GV, et al. KEYNOTE-006
Reck M, DeGreen HP, Rose AL, et al. Avelumab investigators. Pembrolizumab versus ipilimumab in
(MSB0010718C; anti-PD-L1) vs platinum-based doublet as advanced melanoma. N Engl J Med. 2015;372(26):2521-
first-line treatment for metastatic or recurrent PD-L1-positive 2532.
non-small-cell lung cancer: The phase 3 JAVELIN Lung 100 trial.
J Clin Oncol. 2016;34 (suppl):abstr TPS9105. Robert C, Thomas L, Bondarenko I, et al. Ipilimumab plus
dacarbazine for previously untreated metastatic mela-
Reck M, Rodríguez-Abreu D, Robinson AG, et al. KEY- noma. N Engl J Med. 2011;364(26):2517-2526.
NOTE-024 Investigators. Pembrolizumab versus chemo-
therapy for PD-L1-positive non-small-cell lung cancer. Rodriguez PC, Popa X, Martínez O, et al. A phase III clini-
N Engl J Med. 2016;375(19):1823-1833. cal trial of the epidermal growth factor vaccine CIMA-
vax-EGF as switch maintenance therapy in advanced
Rekhtman N, Ang DC, Sima CS, Travis WD, Moreira non-small cell lung cancer patients. Clin Cancer Res.
AL. Immunohistochemical algorithm for differentiation 2016;22(15):3782-3790.
of lung adenocarcinoma and squamous cell carcinoma
based on large series of whole-tissue sections with vali- Rooney MS, Shukla SA, Wu CJ, Getz G, Hacohen N.
dation in small specimens. Mod Pathol. 2011;24(10): Molecular and genetic properties of tumors associated
1348-1359. with local immune cytolytic activity. Cell. 2015;160
(1-2):48-61.
Ribas A, Chow LQ, Boyd JK, et al. Avelumab (MSB0010718C;
anti-PD-L1) in combination with other cancer immunothera- Rosenberg JE, Hoffman-Censits J, Powles T, et al.
pies in patients with advanced malignancies: The phase 1b/2 Atezolizumab in patients with locally advanced and
JAVELIN Medley study. J Clin Oncol. 2016;34(suppl): abstr metastatic urothelial carcinoma who have progressed
TPS3106. following treatment with platinum-based chemother-
apy: A single-arm, multicentre, phase 2 trial. Lancet.
Rimm D, Han G, Taube JM, et al. ORAL01.01: A pro- 2016;387(10031):1909-1920.
spective, multi-institutional assessment of four assays for
PD-L1 expression in NSCLC by immunohistochemistry: Rosenberg SA, Mule JJ, Spiess PJ, Reichert CM, Schwarz
Topic: Pathology. J Thorac Oncol. 2016;11(11S):S249. SL. Regression of established pulmonary metastases and
subcutaneous tumor mediated by the systemic admin-
Rimm DL, Han G, Taube JM, et al. A prospective, multi- istration of high-dose recombinant interleukin 2. J Exp
institutional, pathologist-based assessment of 4 immuno- Med. 1985;161(5):1169-1188.
histochemistry assays for PD-L1 expression in non-small
cell lung cancer. JAMA Oncol. 2017 Mar 9. doi: 10.1001/ Rosenberg SA, Yang JC, Restifo NP. Cancer immuno-
jamaoncol.2017.0013. [Epub ahead of print] therapy: Moving beyond current vaccines. Nat Med.
2004;10(9):909-915.
Rittmeyer A, Barlesi F, Waterkamp D, et al. OAK Study
Group. Atezolizumab versus docetaxel in patients with Rosenberg SA, Yang JC, Sherry RM, et al. Durable com-
previously treated non-small-cell lung cancer (OAK): a plete responses in heavily pretreated patients with meta-
phase 3, open-label, multicentre randomised controlled static melanoma using T-cell transfer immunotherapy.
trial. Lancet. 2017;389(10066):255-265. Clin Cancer Res. 2011;17(13):4550-4557.
Rousseaux S, Debernardi A, Jacquiau B, et al. Ectopic chemotherapy in non-small cell lung cancer. Sci Rep.
activation of germline and placental genes identifies 2016;6:20090.
aggressive metastasis-prone lung cancers. Sci Transl Med.
Sholl LM, Aisner DL, Allen TC, et al. Programmed
2013;5(186):186ra166.
death ligand-1 immunohistochemistry--a new chal-
Rüschoff J, Dietel M, Baretton G, et al. HER2 diagnostics lenge for pathologists: a perspective from members of
in gastric cancer-guideline validation and development the Pulmonary Pathology Society. Arch Pathol Lab Med.
of standardized immunohistochemical testing. Virchows 2016;140(4):341-344.
Arch. 2010;457(3):299-307.
Skov BG, Skov T. P2.01-048 Paired comparison of PDL1
Rüschoff J, Kerr KM, Grote HJ, et al. Reproducibility of assessment on cytology and histology from malignan-
immunohistochemical scoring for epidermal growth factor cies in the lung. J Thorac Oncol. 2017;12(1, suppl):
receptor expression in non-small cell lung cancer: Round S815.
robin test. Arch Pathol Lab Med. 2013;137:1255-1261.
Slamon DJ, Leyland-Jones B, Shak S, et al. Use of che-
Sandler A, Gray R, Perry MC, et al. Paclitaxel-carbopl- motherapy plus a monoclonal antibody against HER2
atin alone or with bevacizumab for non-small-cell lung for metastatic breast cancer that overexpresses HER2.
cancer. N Engl J Med. 2006;355(24):2542-2550. N Engl J Med. 2001;344(11):783-792.
Savic S, Diebold J, Zimmermann AK, et al. Screening for Smith J, Robida MD, Acosta K, et al. Quantitative and
ALK in non-small cell lung carcinomas: 5A4 and D5F3 qualitative characterization of two PD-L1 clones: SP263
antibodies perform equally well, but combined use and E1L3N. Diagn Pathol. 2016;11(1):44.
with FISH is recommended. Lung Cancer. 2015;89(2):
Socinski M, Creelan B, Horn L, et al. CheckMate 026: A
104-109.
phase 3 trial of nivolumab vs investigator’s choice (IC) of
Scheel AH, Dietel M, Heukamp LC, et al. Harmonized platinum-based doublet chemotherapy (PT-DC) as first-
PD-L1 immunohistochemistry for pulmonary squamous- line therapy for stage iv/recurrent programmed death
cell and adenocarcinomas. Mod Pathol. 2016;29(10): ligand 1 (PD-L1)−positive NSCLC. Ann Oncol. 2016;27
1165-1172. (suppl 6): LBA7 PR.
Schiller JH, Harrington D, Belani CP, et al. Comparison of Solomon BJ, Mok T, Kim DW, et al. First-line crizotinib
four chemotherapy regimens for advanced non-small-cell versus chemotherapy in ALK-positive lung cancer.
lung cancer. N Engl J Med. 2002;346(2):92-98. N Engl J Med. 2014;371(23):2167-2177.
Schiller JH, Morgan-Ihrig C, Levitt ML. Concomitant Song Z, Yu X, Cheng G, Zhang Y. Programmed death-
administration of interleukin-2 plus tumor necrosis ligand 1 expression associated with molecular charac-
factor in advanced non-small cell lung cancer. Am J Clin teristics in surgically resected lung adenocarcinoma. J
Oncol. 1995;18(1):47-51. Transl Med. 2016;14(1):188.
Schreiber RD, Old LJ, Smyth MJ. Cancer immunoediting: Spigel DR, Chaft JE, Gettinger SN, et al. Clinical activity
Integrating immunity’s roles in cancer suppression and and safety from a phase II study (FIR) of MPDL3280A
promotion. Science. 2011;331(6024):1565-1570. (anti-PDL1) in PD-L1–selected patients with non-small
cell lung cancer (NSCLC). J Clin Oncol. 2015;33(15_
Schrijver WA, van der Groep P, Hoefnagel LD, et al.
suppl): abstr 8028.
Influence of decalcification procedures on immunohis-
tochemistry and molecular pathology in breast cancer. Stewart R, Morrow M, Hammond SA, et al. Identifica-
Mod Pathol. 2016;29(12):1460-1470. tion and characterization of MEDI4736, an antagonistic
anti-PD-L1 monoclonal antibody. Cancer Immunol Res.
Seiwert TY, Burtness B, Mehra R, et al. Safety and clinical
2015;3(9):1052-1062.
activity of pembrolizumab for treatment of recurrent or
metastatic squamous cell carcinoma of the head and neck Study to Evaluate Safety, Efficacy, Pharmacokinetics and
(KEYNOTE-012): an open-label, multicentre, phase 1b pharmacodynamics of avelumab In combination with
trial. Lancet Oncol. 2016;17(7):956-965. either crizotinib or PF-06463922 in patients with NSCLC.
(Javelin Lung 101)
Shaw AT, Ou SH, Bang YJ, et al. Crizotinib in ROS1-
rearranged non-small-cell lung cancer. N Engl J Med. Sundar R, Soong R, Cho BC, Brahmer JR, Soo RA. Immu-
2014;371(21):1963-1971. notherapy in the treatment of non-small cell lung cancer.
Lung Cancer. 2014 Aug;85(2):101-9.
Sheffield BS, Fulton R, Kalloger SE, et al. Investigation
of PD-L1 biomarker testing methods for PD-1 Axis Sznol M, Chen L. Antagonist antibodies to PD-1 and
inhibition in non-squamous non-small cell lung cancer. B7-H1 (PD-L1) in the treatment of advanced human
J Histochem Cytochem. 2016;64(10):587-600. cancer. Clin Cancer Res. 2013;19(19):1021-1034.
Sheng J, Fang W, Yu J, et al. Expression of programmed Takeda K. Phase II studies of nivolumab in patients
death ligand-1 on tumor cells varies pre and post with advanced squamous (SQ) or non-squamous (NSQ)
REFERENCES 127
non-small-cell lung cancer (NSCLC). Presented at World Pleura, Thymus and Heart. 4th ed. Lyon, France: IARC
Conference on Lung Cancer; September 7, 2015; Press; 2015.
Denver, CO.
Travis WD, Brambilla E, Noguchi M, et al. International
Takada K, Okamoto T, Shoji F, et al. Clinical significance Association for the Study of Lung Cancer/American
of PD-L1 protein expression in surgically resected pri- Thoracic Society/European Respiratory Society
mary lung adenocarcinoma. J Thorac Oncol. 2016;11(11): International Multidisciplinary Classification of Lung
1879-1890. Adenocarcinoma. J Thorac Oncol. 2011;6(2):244-285.
Tan DS, Yom SS, Tsao MS, et al. The International Asso- Tsao MS, Le Teuff G, Shepherd FA, et al. PD-L1 protein
ciation for the Study of Lung Cancer consensus statement expression assessed by immunohistochemistry is
on optimizing management of EGFR mutation-positive neither prognostic nor predictive of benefit from
non-small cell Lung cancer: Status in 2016. J Thorac adjuvant chemotherapy in resected non-small cell
Oncol. 2016;11(7):946-963. lung cancer. Ann Oncol. 2017 Jan 30. [Epub ahead
of print]
Tang Y, Fang W, Zhang Y, et al. The association between
PD-L1 and EGFR status and the prognostic value of Tsao, MS, Hirsch FR, Yatabe Y, eds. IASLC Atlas of ALK
PD-L1 in advanced non-small cell lung cancer patients and ROS1 Testing in Lung Cancer. North Fort Myers,
treated with EGFR-TKIs. Oncotarget. 2015;6(16):14209- FL: Editorial Rx Press; 2016.
142019.
Tumeh PC, Harview CL, Yearley JH, et al. PD-1 block-
Teng MW, Ngiow SF, Ribas A, et al. Classifying cancers ade induces responses by inhibiting adaptive immune
based on T cell infiltration and PD-L1. Cancer Res. resistance. Nature. 2014;515(7528):568-571.
2015;75:2139-2145.
UK NEQAS. http://www.ukneqas.org.uk (accessed
Theiss AP, Chafin D, Bauer DR, Grogan TM, Baird GS. April 3, 2017)
Immunohistochemistry of colorectal cancer biomarker
United States Food and Drug Administration. List of
phosphorylation requires controlled tissue fixation. PLoS
Cleared or Approved Companion Diagnostic Devices
One. 2014;9(11):e113608.
(In Vitro and Imaging Tools. http://www.fda.gov/
Thunnissen E, Borczuk AC, Flieder DB, et al. The use of MedicalDevices/ProductsandMedicalProcedures/
immunohistochemistry improves the diagnosis of small InVitroDiagnostics/ucm301431.htm (accessed August
cell lung cancer and its differential diagnosis. An interna- 13, 2016).
tional reproducibility study in a demanding set of cases.
Uruga H, Bozkurtlar E, Huynh TG, et al. Programmed
J Thorac Oncol. 2017;12(2):334-346.
Cell Death Ligand (PD-L1) Expression in Stage II and
Thunnissen E, Kerr KM, Herth FJ, et al. The challenge III Lung Adenocarcinomas and Nodal Metastases.
of NSCLC diagnosis and predictive analysis on small J Thorac Oncol. 2017;12(3):458-466.
samples. Practical approach of a working group. Lung
US Food and Drug Administration. List of Cleared or
Cancer. 2012;76(1):1-18.
Approved Companion Diagnostic Devices (In Vitro and
Tivol EA, Borriello F, Schweitzer AN, Lynch WP, Blue- Imaging Tools). https://www.fda.gov/medicaldevices/
stone JA, Sharpe AH. Loss of CTLA-4 leads to massive productsandmedicalprocedures/invitrodiagnostics/
lymphoproliferation and fatal multiorgan tissue destruc- ucm301431.htm. Updated February 24, 2017. Accessed
tion, revealing a critical negative regulatory role of March 21, 2017.
CTLA-4. Immunity. 1995;3(5):541-547.
van Seters M, van Beurden M, ten Kate FJ, et al. Treat-
Topalian SL, Hodi FS, Brahmer JR, et al. Safety, activity, ment of vulvar intraepithelial neoplasia with topical
and immune correlates of anti-PD-1 antibody in cancer. imiquimod. N Engl J Med. 2008;358(14):1465-1473.
N Engl J Med. 2012;366(26):2443-2454.
Vansteenkiste JF, Cho BC, Vanakesa T, et al. Efficacy of
Torlakovic EE, Francis G, Garratt J, et al. Standardiza- the MAGE-A3 cancer immunotherapeutic as adjuvant
tion of negative controls in diagnostic immunohisto- therapy in patients with resected MAGE-A3-positive
chemistry: recommendations from the international non-small-cell lung cancer (MAGRIT): a randomised,
ad hoc expert panel. Appl Immunohistochem Mol double-blind, placebo-controlled, phase 3 trial. Lancet
Morphol. 2014;22(4):241-252. Oncol. 2016;17(6):822-835.
Torlakovic EE, Nielsen S, Francis G, et al. Standardiza- Vassilakopoulou M, Parisi F, Siddiqui S, et al. Preanalyt-
tion of positive controls in diagnostic immunohisto- ical variables and phosphoepitope expression in FFPE
chemistry: recommendations from the International tissue: quantitative epitope assessment after variable
Ad Hoc Expert Committee. Appl Immunohistochem Mol cold ischemic time. Lab Invest. 2015;95(3):334-341.
Morphol. 2015;23(1):1-18.
Velcheti V, Schalper KA, Carvajal DE, et al. Pro-
Travis WD, Brambilla E, Burke AP, Marx A, Nicholson grammed death ligand-1 expression in non-small cell
AG, eds. WHO Classification of Tumours of the Lung, lung cancer. Lab Invest. 2014;94(1):107-116.
Ventana Medical Systems, Inc. Ventana PD-L1 (SP142) Wu YL, Park K, Soo RA, et al. INSPIRE: A phase III
Assay [package insert]. Accessed March 15, 2017. study of the BLP25 liposome vaccine (L-BLP25) in Asian
patients with unresectable stage III non-small cell lung
Ventana Medical Systems, Inc. Ventana PD-L1 (SP142)
cancer. BMC Cancer. 2011;11:430.
Assay [package insert]. http://www.ventana.com/
product/1815?type=2324. Accessed April 4, 2017. Yang CY, Lin MW, Chang YL, Wu CT, Yang PC.
Programmed cell death-ligand 1 expression in surgi-
Ventana Medical Systems, Inc. Ventana PD-L1 (SP142)
cally resected stage I pulmonary adenocarcinoma and its
Assay Interpretation Guide for Non-Small Cell Lung
correlation with driver mutations and clinical outcomes.
Cancer. Accessed March 15, 2017.
Eur J Cancer. 2014;50(7):1361-1369.
Verschraegen CF, Chen F, Spigel DR, et al. Avelumab
Yang CY, Lin MW, Chang YL, Wu CT, Yang PC. Pro-
(MSB0010718C; anti-PD-L1) as a first-line treatment for
grammed cell death-ligand 1 expression is associated
patients with advanced NSCLC from the JAVELIN solid
with a favourable immune microenvironment and better
tumor phase 1b trial: safety, clinical activity, and PD-L1
overall survival in stage I pulmonary squamous cell
expression. J Clin Oncol. 2016;34(suppl): abstr 9036.
carcinoma. Eur J Cancer. 2016;57:91-103.
Viard-Leveugle I, Veyrenc S, French LE, Brambilla C,
Yeh YC, Lin SF, Chiu CH, et al. PD-L1 status in
Brambilla E. Frequent loss of Fas expression and function
Taiwanese lung adenocarcinoma patients: Comparison
in human lung tumours with overexpression of FasL in
of PD-L1 immunohistochemical assays using antibody
small cell lung carcinoma. J Pathol. 2003;201(2):268-277.
clones 22C3, SP142 and SP263 with clinicopathological
Vyberg M, Nielsen S. Proficiency testing in immuno- correlation. Ann Oncol. 2016;27(suppl_9): mdw588004.
histochemistry--Experiences from Nordic Immuno-
Yu H, Boyle TA, Zhou C, Rimm DL, Hirsch FR.
histochemical Quality Control (NordiQC). Virchows
PD-L1 expression in lung cancer. J Thorac Oncol. 2016;
Arch. 2016;468(1):19-29.
11(7):964-975.
Wang C, Thudium KB, Han M, et al. In vitro characteriza-
Zaretsky J. High intratumoral T cell -infiltration corre-
tion of the anti-PD-1 antibody nivolumab, BMS-936558,
lated with mutational load and response to pembroli-
and in vivo toxicology in non-human primates. Cancer
zumab in non-small cell lung cancer. Presented at World
Immunol Res. 2014;2(9):846-856.
Conference on Lung Cancer, September 9, 2015; Denver,
Waterhouse P, Penninger JM, Timms E, et al. Lymphopro- CO. Abstract ORAL 31.05.
liferative disorders with early lethality in mice deficient
Zhang Y, Wang L, Li Y, et al. Protein expression of
in Ctla-4. Science. 1995;270(5238):985-988.
programmed death 1 ligand 1 and ligand 2 independently
World Health Organization International Agency for predict poor prognosis in surgically resected lung
Research on Cancer. GLOBOCAN2012: Estimated Cancer adenocarcinoma. Onco Targets Ther. 2014;7:567-573.
Incidence, Mortality and Prevalence Worldwide in 2012.
http://globocan.iarc.fr/Pages/fact_sheets_cancer.aspx.
Accessed March 8, 2017.
The IASLC Atlas of PD-L1 Immunohistochemistry Testing in Lung Cancer
is a resource designed to help pathologists, clinicians, other health care
personnel, and patients to better understand emerging programmed
cell death ligand-1 (PD-L1) immunohistochemistry (IHC) assays as well
as important areas of clarity and debate.
At present, although PD-L1 protein expression, as detected by IHC

testing, is widely used as a predictive biomarker assay for anti–PD-1/
PD-L1 therapies, more information is needed regarding interpretation,
assay usage for PD-L1 testing, and potential interchangeability. The
editors and authors provide this additional information by looking at
the changing landscape of laboratory testing, the specifics of each
assay, and the current controversies regarding PD-L1 expression testing
in lung cancer.
It is IASLC’s goal that through the creation of this Atlas, patients

with lung cancer will receive the most contemporary and well-suited
treatment options, based on up-to-date evidence, and will feel more
confident and knowledgeable regarding their therapy.
IASLC acknowledges the generous funding and support provided

by AstraZeneca, Bristol-Myers Squibb, and Merck for the IASLC
Atlas of PD-L1 Immunohistochemistry Testing in Lung Cancer.
North Fort Myers, FL

www.EditorialRxPress.com

PkljKYwv PDL1 Atlas 2016 PDF

Uploaded by

Copyright:

Available Formats

PkljKYwv PDL1 Atlas 2016 PDF

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

PkljKYwv PDL1 Atlas 2016 PDF

Uploaded by

Copyright:

Available Formats

INTERNATIONAL ASSOCIATION FOR THE STUDY OF LUNG CANCER

IASLC ATLAS OF PD-L1

An IASLC publication published by Editorial Rx Press

Cover and interior design by Amy Boches, Biographics

April 2017, October 2017

Copyright © 2017 International Association for the Study of Lung Cancer

While the information in this book is believed to be true and accurate as of

Erratum: The images in Figures 8 and 10 in Chapter 3, “Immunohistochemistry

AN INTERNATIONAL ASSOCIATION FOR THE STUDY OF LUNG CANCER PUBLICATION

IASLC acknowledges the generous funding and support

8 PD-L1 73-10 Assay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79

Editors Contributing Authors

Yasushi Yatabe, MD, PhD Teh-Ying Chou, MD, PhD

Sylvie Lantuéjoul, MD, PhD Ross A. Soo, MBBS

Left to right: Teh-Ying Chou, Yasushi Yatabe, Lukas

Abcam Leica Biosystems

Agilent Technologies/Dako Proteintech Group, Inc.

Cancer and T-Cells

T cells recognize an antigen, which is Box 1. Major Immunotherapeutic Approaches

followed by expansion with interleukin-2 (IL-2) stimulation and reinfusion immediately

Coordination of Immune Responses

Cytokines Stimulating Antigen-presenting cell T cell

T cells have the ability to directly CD80 or CD86 CTLA-4 –

ecules. In contrast, CD4 T cells

recognize peptides presented by HVEM BTLA –

MHC class II molecules and mediate KIR –

cells) for suppression of immunity. Adenosine

years, suggesting remarkable durability of anticancer responses in contrast to other cyto-

Box 2. Mechanisms of Escape from Immune Surveillance

PD-1 and PD-L1

cells express PD-L1 to escape from immune

Anergy of CD8 cytotoxicity.

Counter-regulation by excessive CD25+FOXP3+CD4+ on CD8+PD-1+ Treg cells in the immune

Unavailability of functional apoptotic pathways in tumor cells by lack of functional Fas-FasL

Antigen Antigen Anti-tumor Adoptive cell

Therapeutic Modulating T Enhancing immune

GSK1572932A Immune checkpoint Cytokines: TCR engineering

In the advanced-disease setting, TG4010, another MUC1-targeting vaccine that uses

Immune Checkpoint Inhibitors

with placebo plus dacarbazine (18.2% [95% Activation

prior treatment. In this three-arm phase II

0.72; p=0.05) and 5.5 months for the concur-

ated for patients with squamous cell NSCLC.

PD-1 Inhibitors Figure 2. The anti-cytotoxic T lymphocyte-associated pro-

Activated CTL recently, the combination of nivolumab and ipi-

Trial Treatment Histologic ORR Progression-free

Borghaei Phase Second Nonsq Nivolumab 19 2.3 NR 19 12.2 (9.7-15.0) 51

continued on next page

Barlesi Phase Second or Sq and Atezolizumab 14 2.8 NR NR 13.8 55

Atezolizumab was reported to be active in a phase I study of advanced-stage NSCLC

Combined CTLA-4 and PD-1/PD-L1 Inhibitors

Detection of PD-L1 Expression

Durvalumab SP263 (rabbit) Cytoplasmica,b BenchMark ≥25% tumor cells

the intensity level, as first applied to 20

approach may lead to more unifor-

samples because final results also

PD-L1 Validation Table 3. Comparison of Analytic and Clinical Validation in

including preanalytic, analytic, and Requires treatment outcome in No Yes

PD-L1 Interpretation “Histology”

Examination of the Patient Slide