PkljKYwv PDL1 Atlas 2016 PDF
PkljKYwv PDL1 Atlas 2016 PDF
PkljKYwv PDL1 Atlas 2016 PDF
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
IASLC Office:
IASLC, 13100 East Colfax Ave., Unit 10, Aurora, Colorado 80011, USA
www.iaslc.org
ISBN: 978-0-9832958-7-7
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.
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
Editorial Rx Press
North Fort Myers, FL
Acknowledgments
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
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.
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
12 IASLC ATLAS OF EGFR TESTING IN LUNG CANCER
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).
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
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).
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
18 IASLC ATLAS OF PD-L1 IMMUNOHISTOCHEMISTRY TESTING IN LUNG CANCER
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.
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
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
20 IASLC ATLAS OF PD-L1 IMMUNOHISTOCHEMISTRY TESTING IN LUNG CANCER
lung cancer (Brambilla 2016), and low density of active CD8+/PD-1+ CTLs has been reported
(Tumeh 2014, Kim 2015).
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
22 IASLC ATLAS OF PD-L1 IMMUNOHISTOCHEMISTRY TESTING IN LUNG CANCER
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
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).
24 IASLC ATLAS OF PD-L1 IMMUNOHISTOCHEMISTRY TESTING IN LUNG CANCER
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
26 IASLC ATLAS OF PD-L1 IMMUNOHISTOCHEMISTRY TESTING IN LUNG CANCER
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
CANCER IMMUNOTHERAPY FOR LUNG CANCER 27
Table 2. Clinical Outcomes for Patients with Advanced NSCLC Treated with Anti–PD-1/PD-L1 Immune
Checkpoint Inhibitors
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
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
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
30 IASLC ATLAS OF PD-L1 IMMUNOHISTOCHEMISTRY TESTING IN LUNG CANCER
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
CANCER IMMUNOTHERAPY FOR LUNG CANCER 31
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).
32 IASLC ATLAS OF PD-L1 IMMUNOHISTOCHEMISTRY TESTING IN LUNG CANCER
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.
CANCER IMMUNOTHERAPY FOR LUNG CANCER 33
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).
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),
36 IASLC ATLAS OF PD-L1 IMMUNOHISTOCHEMISTRY TESTING IN LUNG CANCER
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
38 IASLC ATLAS OF PD-L1 IMMUNOHISTOCHEMISTRY TESTING IN LUNG CANCER
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
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
IMMUNOHISTOCHEMISTRY FOR PD-L1 39
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.)
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.
40 IASLC ATLAS OF PD-L1 IMMUNOHISTOCHEMISTRY TESTING IN LUNG CANCER
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).
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.
has weak epitope concentration, allowing detection of minor deviations in the staining
protocol.
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).
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
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.
IMMUNOHISTOCHEMISTRY FOR PD-L1 45
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.
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.
IMMUNOHISTOCHEMISTRY FOR PD-L1 47
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.
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.
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.
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).
52 IASLC ATLAS OF PD-L1 IMMUNOHISTOCHEMISTRY TESTING IN LUNG CANCER
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.
54 IASLC ATLAS OF PD-L1 IMMUNOHISTOCHEMISTRY TESTING IN LUNG CANCER
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).
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
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,
58 IASLC ATLAS OF PD-L1 IMMUNOHISTOCHEMISTRY TESTING IN LUNG CANCER
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
60 IASLC ATLAS OF PD-L1 IMMUNOHISTOCHEMISTRY TESTING IN LUNG CANCER
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
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
62 IASLC ATLAS OF PD-L1 IMMUNOHISTOCHEMISTRY TESTING IN LUNG CANCER
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
64 IASLC ATLAS OF PD-L1 IMMUNOHISTOCHEMISTRY TESTING IN LUNG CANCER
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.
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.
TC < 50%
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
*Number of cases evaluable for PD-L1 expression. NSCLC = non-small cell lung cancer, NA = not available
PD-L1 SP142 ASSAY
67
68 IASLC ATLAS OF PD-L1 IMMUNOHISTOCHEMISTRY TESTING IN LUNG CANCER
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.
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
PD-L1 SP142 ASSAY 69
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.
70 IASLC ATLAS OF PD-L1 IMMUNOHISTOCHEMISTRY TESTING IN LUNG CANCER
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
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.
72 IASLC ATLAS OF PD-L1 IMMUNOHISTOCHEMISTRY TESTING IN LUNG CANCER
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
74 IASLC ATLAS OF PD-L1 IMMUNOHISTOCHEMISTRY TESTING IN LUNG CANCER
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+).
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.
PD-L1 SP263 ASSAY 75
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).
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
PD-L1 SP263 ASSAY 77
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).
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).
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.
82 IASLC ATLAS OF PD-L1 IMMUNOHISTOCHEMISTRY TESTING IN LUNG CANCER
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.
ORR = objective response rate, PD-L1 = programmed cell death ligand-1, HR = hazard ratio, NA = not applicable.
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
84 IASLC ATLAS OF PD-L1 IMMUNOHISTOCHEMISTRY TESTING IN LUNG CANCER
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).
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
(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.
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).
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)
*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.
OTHER ANTI–PD-L1 CLONES: ALTERNATIVE ASSAYS AND LABORATORY-DEVELOPED TESTS 89
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.
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.
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.
94 IASLC ATLAS OF PD-L1 IMMUNOHISTOCHEMISTRY TESTING IN LUNG CANCER
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
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
98 IASLC ATLAS OF PD-L1 IMMUNOHISTOCHEMISTRY TESTING IN LUNG CANCER
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.
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
ASSAY HARMONIZATION: IS IT POSSIBLE? 101
we have (in the absence of validating clinical outcome data) is comparison with at least one
of the trial-validated commercial assays.
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).
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.
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
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..
104 IASLC ATLAS OF PD-L1 IMMUNOHISTOCHEMISTRY TESTING IN LUNG CANCER
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)
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
ASSAY HARMONIZATION: IS IT POSSIBLE? 105
100
22C3
90 28-8
SP142
80 SP263
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
106 IASLC ATLAS OF PD-L1 IMMUNOHISTOCHEMISTRY TESTING IN LUNG CANCER
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.
Table 3. Potential Immunohistochemistry (IHC) Testing Choices for Determination of Programmed Cell
Death Ligand-1 (PD-L1) Expression
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.
ASSAY HARMONIZATION: IS IT POSSIBLE? 107
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).
110 IASLC ATLAS OF PD-L1 IMMUNOHISTOCHEMISTRY TESTING IN LUNG CANCER
Resection Biopsy
Core/LN Cytology
sampling samples
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.
unstained sections
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.
112 IASLC ATLAS OF PD-L1 IMMUNOHISTOCHEMISTRY TESTING IN LUNG CANCER
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.
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–
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’
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
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
118 IASLC ATLAS OF PD-L1 IMMUNOHISTOCHEMISTRY TESTING IN LUNG CANCER
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
120 IASLC ATLAS OF PD-L1 IMMUNOHISTOCHEMISTRY TESTING IN LUNG CANCER
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
122 IASLC ATLAS OF PD-L1 IMMUNOHISTOCHEMISTRY TESTING IN LUNG CANCER
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
124 IASLC ATLAS OF PD-L1 IMMUNOHISTOCHEMISTRY TESTING IN LUNG CANCER
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.
126 IASLC ATLAS OF PD-L1 IMMUNOHISTOCHEMISTRY TESTING IN LUNG CANCER
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.
128 IASLC ATLAS OF PD-L1 IMMUNOHISTOCHEMISTRY TESTING IN LUNG CANCER
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.