-
Cancer Therapy: Preclinical
PD-L1 Studies Across Tumor Types, Its DifferentialExpression and
Predictive Value in PatientsTreated with Immune Checkpoint
InhibitorsHarriet M. Kluger1, Christopher R. Zito2,Gabriela
Turcu1,3, Marina K. Baine4, Hongyi Zhang1,Adebowale Adeniran4,
Mario Sznol1, David L. Rimm4, Yuval Kluger4, Lieping Chen5,Justine
V. Cohen1, and Lucia B. Jilaveanu1
Abstract
Purpose: With recent approval of inhibitors of PD-1 in
mel-anoma, non–small cell lung cancer (NSCLC) and renal
cellcarcinoma, extensive efforts are under way to develop
biomarkerspredictive of response. PD-L1 expression has been most
widelystudied, and is more predictive in NSCLC than renal cell
carci-noma or melanoma. We therefore studied differences in
expres-sion patterns across tumor types.
Experimental Design:Weused tissuemicroarrays with
tumorsfromNSCLC,renalcell carcinoma,ormelanomaandapanelof celllines
to study differences between tumor types. Predictive studieswere
conducted on samples from 65 melanoma patients treatedwith PD-1
inhibitors alone or with CTLA-4 inhibitors, character-ized for
outcome. PD-L1 expression was studied by
quantitativeimmunofluorescence using two well-validated
antibodies.
Results: PD-L1 expression was higher in NSCLC specimensthan
renal cell carcinoma, and lowest inmelanoma (P¼ 0.001),
and this finding was confirmed in a panel of cell lines.
Inmelanoma tumors, PD-L1 was expressed either on tumorcells or
immune-infiltrating cells. The association betweenPD-L1 expression
in immune-infiltrating cells and progres-sion-free or
overall-survival in melanoma patients treated withipilimumab and
nivolumab was stronger than PD-L1 expres-sion in tumor cells, and
remained significant on multivariableanalysis.
Conclusions: PD-L1 expression in melanoma tumor cellsis lower
than NSCLC or renal cell carcinoma cells. The higherresponse rate
in melanoma patients treated with PD-1 inhibi-tors is likely
related to PD-L1 in tumor-associated inflammatorycells. Further
studies are warranted to validate the predictiverole of
inflammatory cell PD-L1 expression in melanoma anddetermine its
biological significance. Clin Cancer Res; 23(15);4270–9. �2017
AACR.
IntroductionImmune checkpoint inhibitors have become the
mainstay of
treatment formelanoma and other tumor types. The first
immunecheckpoint inhibitor to gain approval, ipilimumab,
inhibitsCTLA-4 on cytotoxic T cells, resulting in durable responses
in11% to19%ofpatientswith advancedmelanomaandprolongingoverall
survival (OS; refs. 1–3). Treatment with ipilimumab,however, causes
grade 3 to 4 immune-related adverse events in
approximately 30% of patients at the FDA-approved dose of
3mg/kg, diminishing the risk–benefit ratio of this drug.
Inhibitors of PD-1 or its ligand, PD-L1, have similarly
beenstudied in advanced melanoma and other tumor types, and havenow
been approved for a number of diseases including melano-ma, renal
cell carcinoma, bladder cancer, non–small cell lungcancer (NSCLC),
head and neck cancer, and Hodgkin's lympho-ma (4–12). Response
rates to PD-1 and PD-L1 inhibitors inmelanomawere higher than those
of ipilimumab, and the toxicityprofile more favorable, with
response rates in the range of 30% to40% and approximately 15% of
patients having grade 3 to 4immune-related adverse events (1,
4–6).
The combination of ipilimumab and nivolumab has beenstudied in a
number of diseases, and is now approved foradvanced melanoma. The
response rate with the combinationwas superior to that of either
drug alone (57.6% in the first linesetting), and the rate of grade
3 to 4 adverse events was 55%,morethan double that of monotherapy
(1, 13, 14). Biomarkers pre-dictive of response or resistance are
therefore needed to improvepatient selection, and given that this
is a relatively new regimenwith limited patient follow-up,
predictive biomarkers have barelybeen studied.
To date, despite a number of attempts to identify
biomarkerspredictive of response to ipilimumabmonotherapy, no
biomark-er has consistently been shown to be associated with
responseor clinical benefit (15, 16). Given the broader use of
inhibitors of
1Department of Medicine, Yale University School of Medicine, New
Haven,Connecticut. 2Department of Biology, School of Health and
Natural Sciences,University of Saint Joseph, West Hartford,
Connecticut. 3Department of Der-matology, Carol Davila University
of Medicine and Pharmacy, Bucharest, Roma-nia. 4Department of
Pathology, Yale University School of Medicine, New
Haven,Connecticut. 5Department of Immunobiology, Yale University
School of Med-icine, New Haven, Connecticut.
H.M. Kluger and C.R. Zito contributed equally to this
article.
Note: Supplementary data for this article are available at
Clinical CancerResearch Online
(http://clincancerres.aacrjournals.org/).
Corresponding Author: Harriet M. Kluger, Yale University School
of Medicine,333 Cedar Street, PO BOX 208028, New Haven, CT 06520.
Phone: 203-737-2572, Fax: 203-785-3788; E-mail:
[email protected]
doi: 10.1158/1078-0432.CCR-16-3146
�2017 American Association for Cancer Research.
ClinicalCancerResearch
Clin Cancer Res; 23(15) August 1, 20174270
on March 29, 2021. © 2017 American Association for Cancer
Research. clincancerres.aacrjournals.org Downloaded from
Published OnlineFirst February 21, 2017; DOI:
10.1158/1078-0432.CCR-16-3146
http://crossmark.crossref.org/dialog/?doi=10.1158/1078-0432.CCR-16-3146&domain=pdf&date_stamp=2017-7-11http://clincancerres.aacrjournals.org/
-
PD-1 or PD-L1 in multiple tumor types, intense efforts are
underway to identify predictors of response. Expression of PD-L1
ontumor cells has been themost widely studied predictive
biomark-er, and has been shown to correlate with response to
therapy inmultiple tumor types, although the correlation is
insufficient inmost tumor types, includingmelanoma and renal cell
carcinoma,for clinical use. Other predictive biomarkers that have
beenstudied in melanoma tumors include tumor mutation burden,T-cell
receptor repertoire, T-cell infiltrate, gene expression
profiles,and presence of MHC molecules. Inflammatory gene
expressionsignatures within the tumor, particularly those
associated withIFNg secretion, are associated with response to PD-1
inhibitors(17). Tumors with a greater mutation load might be
moresensitive, particularly BRCA2 mutations (18). Presence of
CD8þ
T cells at the periphery of the melanoma tumor bed is
associatedwith a greater likelihood of response to PD-1 inhibitors,
as ispresence of tumor specificMHCclass IImolecules (19, 20).
PD-L1expression, however, is the one biomarker that has
consistentlybeen shown to be associated with response in multiple
trials andclinical settings, albeit insufficiently correlated to be
broadly usedalone as a companion diagnostic.
Most predictive biomarker studies involving PD-L1 expressionhave
used standard IHC, as reviewed (21–23). These studies haveused a
variety of antibodies and cutpoints for positivity (24).
Forexample, in the randomized trial of nivolumab versus
chemo-therapy in the second line setting, 43.6% of patients with
>5%tumor cell staining for PD-L1 had a response, compared to
20.3%of those with
-
ipilimumab and nivolumab and 52% received anti-PD-1
mono-therapy. RECIST 1.1 criteria were used for clinical assessment
andclassification of response.
To assess antibody specificity and for comparison of
stainingacross diseases we used control arrays containing placental
andtonsil tissue (known to be positive for PD-L1 expression),
pelletsfrom MEL-624 cell lines, overexpressing or not
overexpressingPD-L1, and cases of metastatic melanoma, NSCLC and
clear cellrenal cell carcinoma, as previously described (27).
Immunofluorescence and Automated Quantitative AnalysisStaining
was performed for automated analysis, as previously
described (27, 29, 33). Briefly, TMA slideswere deparaffinized
andhydrated followed by antigen retrieval. For the E1L3N
antibody(catalogno. 13684;Cell Signaling Technology), slideswere
boiledfor 20 minutes in a pressure cooker containing
6.5mmol/Lsodium citrate (pH 6.0). Endogenous peroxidase activity
wasblocked using hydrogen peroxide solution. Unspecific stainingwas
blocked in 0.3% bovine serum albumin solution beforeovernight
incubation with anti-PDL1 antibody diluted in TBSsupplementedwith
0.05%Tween20 (1:250, at 4�C). Todetect thecell membrane/cytoplasmic
compartment and to create a tumormask, we simultaneously used a
cocktail of mouse anti-S100 andanti-HMB45 antibodies, diluted at
1:100 in the overnight solu-tion (BioGenex) for melanoma samples.
The tumor mask forNSCLC and renal cell carcinoma was generated
usingmouse anti-cytokeratin (catalog no. M3515; DAKO) or antibody
cocktail ofanti-cytokeratin, CAIX (gift from Jan Zavada), CD10
(catalog no.M7308;DAKO), and streptavidinHRP (catalog no. S2438;
Sigma-Aldrich), respectively. Slides were then incubated with goat
anti-mouse IgG conjugated to Alexa Flour 546 diluted at
1:200(Molecular Probes, Inc.) in anti-rabbit amplification
reagent(Envision; catalog no. K4003; DAKO).
For clone 5H1 (generated by Dr. Lieping Chen), antigenretrieval
was performed in Tris-EDTA buffer (pH 9.0; DAKO)supplemented with
0.3% Tween 20 (34). Following peroxidaseblocking, slides were
incubated at room temperature for 15minutes in ACE blocking buffer.
To block endogenous biotin,slides were first incubated in Avidin
solution followed by biotinblocking reagent (Vector Laboratories)
for 15 minutes at 37�C.Slides were incubated overnight with clone
5H1 (1:400 at 4�C)and rabbit anti-S100 (1:100, catalog no. Z0311;
DAKO) in ACEblock. A second TMA control slide, incubated overnight
in ACEblock only was used to verify lack of false-positive
staining. Abiotynilated anti-mouse IgG (Vector laboratories) was
used as asecondary reagent (1:500 for 45 minutes at room
temperature),followed by incubationwith the ABCKit (Vector
Laboratories) for15 minutes. Amplification reagent and
streptavidin-HRP wereused for signal amplification as recommended
(DAKO). Forvisualization of S100 slides were incubated with Alexa
Fluor546-conjugated anti-rabbit IgG.
For T-cell detection, the following primary antibodies
wereutilized: rabbit anti-CD3 (1: 3200, catalog no. A0452;
DAKO),mouse anti-CD4 (1:400, catalog no.M7310; DAKO),mouse anti-CD8
(1:200, catalog no. M7103; DAKO), andmouse anti-FOXP3(1:200,
catalog no. ab20034; Abcam). Staining was carried outusing our
standardprotocol. To amplify the signal goat anti-rabbitand/or
anti-mouse Envision were used followed by fluorophoretyramides.
Slides were incubatedwith benzoic hydrazide solution(100 mmol/L in
PBS) containing 50 mmol/L hydrogen peroxidebefore second
amplification to quench residual HRP.
Target stainingwas visualizedwith cyanine-5-tyramide (PD-L1,CD4
and CD8, or cyanine-3-tyramide (CD4 and FOXP3) (Perkin-Elmer). A
nuclear mask was created by incubating the slides
with4,6-diamidine-2-phenylindole (DAPI; 1:500; Invitrogen).
Cover-slips were mounted with ProLong Gold antifade medium
(Invi-trogen/Life Technologies).
Quantitative Determination of Target ExpressionImage capturing
and quantitative measurements were con-
ducted using methods and algorithms previously described(33).
Tumor was distinguished from the surrounding stromalelements by
S-100 signal for melanoma and cytokeratin forNSCLC or renal cell
carcinoma. The tumor mask was created byconverting the anti-S100
image via automated processing andthresholding. DAPI signal
defining the nuclear compartment, wasutilized to create a total
tissue mask (tumor and stroma). Thestromal compartment was obtained
by subtracting the tumormask from the total tissue mask.
Quantification of PD-L1 signalfor each antibody in the tumor or
stromal compartment (totalsignal intensity/area of the compartment)
was performed, andoutput obtained on a scale of 0 to 255. To assess
the degree oftumor-infiltrating lymphocytes (TIL), we used the
percentage ofeither CD3-, CD4- or CD8-, FOXP3þ T-cell area within
the histo-spot. Tumor spots were excluded if they contained
insufficienttissue (
-
5H1, generated by Dr. Lieping Chen and validated in
previouspublications (25, 27, 29), which binds the extracellular
domain ofPD-L1, and a rabbit monoclonal antibody (clone E1L3N
fromCell Signaling), which recognizes the intracellular domain or
totalendogenous level of the protein. To verify antibody
specificity andsensitivity, we first used quantitative
immunofluorescent (QIF)analysis of control TMAs containing
placenta, tonsil tissue(known to be positive for PD-L1 expression),
and pellets fromMEL-624 cell lines, overexpressing or not
overexpressing PD-L1,as previously described (27). Control arrays
also included12 casesof metastatic melanoma, 12 NSCLC, and 21 renal
cell carcinomaspecimens, which allowed comparison of staining
patterns andintensity across the three diseases. Staining was
confined either tothe cell surface (membranous staining) and in
some cases to thecytoplasm and was specific for both antibodies as
judged by ourpositive/negative cell line controls (Figure 1). As
expected, PD-L1was expressed at high levels inMEL624PD-L1þ cells
but not in theparental MEL624 cells. In melanoma specimens, PD-L1
wasexpressed either on tumor cells or in the
tumormicroenvironmenton immune-infiltrating cells. In contrast, in
NSCLC and renal cellcarcinoma specimens from our control arrays,
PD-L1 expressionwas predominantly on tumor cells. An example of a
PD-L1þ casefor each tumor type is shown in Figure 1, corresponding
to thestrongest expressing histospots for each of the three
diseases. Forcomparative studies across tumor types, we quantified
the fluo-rescent signal (either maximum or mean intensity values)
withinthe entire histospot. Expression levels in the corresponding
speci-mens by the two antibodies (5H1 and E1L3N) was highly
cor-related (R¼ 0.89). The expression of PD-L1 inNSCLC
tumorswassignificantly higher by ANOVA when compared to renal
cellcarcinoma and melanoma specimens (mean AQUA score of21.5 vs. 16
vs. 4, respectively for clone 5H1, P ¼ 0.001; 21.5 vs.12.2 vs. 9.6,
respectively for clone E1L3N, P ¼ 0.0008; Supple-mentary Figure S1A
and B).
To further verify differential PD-L1 expression in tumor
cellsfrom the three histologies as opposed to stromal and
immuneinfiltrating cells, we probed lysates from a panel of
tumor-derived melanoma cultures, and NSCLC or renal cell
carcinomacell lines by Western blot analysis (Figure 2). Both
antibodieswere highly specific for their target, demonstrating a
band thatcorresponded to a molecular weight of 48 kDa. Equal
loadingand protein content among samples was confirmed by
b-actinsignal. PD-L1 levels varied significantly by tumor cell
type.Expression levels were high or moderately high in six of
tenNSCLC and two of six renal cell carcinoma lysates tested.
Bycomparison, only two of the nine melanoma cell lines
dem-onstrated PD-L1 expression and levels in these samples weremuch
lower than in NSCLC and renal cell carcinoma lysates.For
semiquantitative assessment of protein levels, signal wasnormalized
to b-actin and fold increase above backgroundsignal was determined
for each sample using ImageJ software.Expression levels in the
corresponding cell lines by the twoantibodies (5H1 and E1L3N) was
highly correlated (R ¼ 0.94).NSCLC cell lines had significantly
higher PD-L1 levels thanrenal cell carcinoma or melanoma lysates by
ANOVA (meanlevels: 14,162 vs. 5,905 vs. 3,032, respectively for
clone 5H1,P ¼ 0.047; 21,612 vs. 11,158 vs. 1,828, respectively for
cloneE1L3N, P ¼ 0.014; Supplementary Figure S1C and S1D).
Association between PD-L1 expression in melanoma tumorversus
tumor microenvironment cells and response toPD-1-based therapy
To assess the association between PD-L1 expression on thecell
surface of either tumor cells (melanoma cells) or
infiltratingimmune cells (stromal cells) and the likelihood of
response toPD-1 blockade (alone or in combination with ipilimumab),
weconstructed a TMA containing three cores from different areasof
pretreatment tumors from 65 melanoma patients. The
Figure 1.
Examples of strong PD-L1 fluorescent staining. Staining is shown
in paraffin-embedded pellets of MEL624 transfected to overexpress
PD-L1 or parental cells,NSCLC, renal cell carcinoma, and melanoma
specimens. Either an anti-S100 antibody or anti-cytokeratin (CK)
antibody and fluorophore tyramidewere used to distinguish tumor
cells (green) from the surrounding stroma and leukocytes. 5H1
antibody (top) or E1L3N antibody (bottom) andCyanine-5-tyramide
(red) was used to visualize PD-L1. Overlaid images show PD-L1
expression within the entire histospot comprising both the
tumor(orange/yellow) and stroma (red), at 10� magnification.
PD-L1 in the Tumor Microenvironment
www.aacrjournals.org Clin Cancer Res; 23(15) August 1, 2017
4273
on March 29, 2021. © 2017 American Association for Cancer
Research. clincancerres.aacrjournals.org Downloaded from
Published OnlineFirst February 21, 2017; DOI:
10.1158/1078-0432.CCR-16-3146
http://clincancerres.aacrjournals.org/
-
diameter of the histocores is similar to that of biopsies,
ren-dering TMAs a useful surrogate for clinical evaluation of
met-astatic samples, which is often based on biopsy rather
thanmetastatectomy. The TMA was stained with either 5H1 orE1L3N
antibodies and immunofluorescence scores were gen-erated using
methods previously described (29). Each variablewas measured
separately in either the tumor or stromal com-partment for the two
antibodies.
For patients who had two or three assessable histospots,
acomposite score was formed by calculating the mean value. Themean
value was compared to themaximum score in these cases todetermine
whether there were significant differences, and wefound a very
strong correlation between mean and maximumvalues across
measurements, with correlation coefficients rangingfrom0.94 to0.98;
Supplementary Figure S2). Therefore, the rest ofthe analyses are
based onmean values only. For patients who hadonly one histospot, a
single value was used.
To assess the variability of PD-L1 staining between clones
5H1andE1L3N, themeanQIF scores (mean intensity eitherwithin
thetumor compartment or stroma) for each case from the
corre-sponding TMAswere compared and found to be highly
correlated(R ¼ 0.86 and R ¼ 0.78, respectively; Supplementary
Figure S3).
To assess the association between PD-L1 expression andresponse
to therapy, we compared PD-L1 expression in twopatient groups:
responders (complete response and partialresponse) and
nonresponders (stable disease and progressivedisease). Our analysis
in the group of patients receiving mono-therapy had a number of
limitations; several were treated withpembrolizumab in an expanded
access trial, some had receivedprior nivolumab, and most had
received multiple other lines oftherapy, and the response rate was
therefore lower than expected.We note that 11 of 31 cases with
PD-L1 scores available foranalysis were treated with ipilimumab
within 3months of receiv-ing anti-PD-1 monotherapy and were
excluded from the analysisdue to the potential interaction between
the drugs. In the smallergroup of 20 remaining patients treated
with PD-1 inhibitormonotherapy, high PD-L1 expression in the
stromal compart-ment was weakly associated with response to
therapy, but this didnot reach statistical significance (P ¼ 0.07
by Chi-square analysisof scores dichotomized by the median value,
data not shown).Given the limitations of this cohort, both in terms
of the lower
than expected response rate and the small cohort size, the
remain-der of the analysis focuses on patients treated with the
combina-tion of ipilimumab and nivolumab.
In patients treated with combination therapy (ipilimumab
andnivolumab), the time between tissue acquisition and initiation
ofimmune checkpoint inhibitors ranged from 1 to 34
months(median-10, mean–11). Twenty-two patients received no
inter-current systemic therapies, six received one line of prior
therapyand three received two prior regimens. Prior therapies
includedIL2 (five patients), IFN (one patient), vemurafenib
(threepatients), dacarbazine, carboplatin plus paclitaxel, and IL21
plusnivolumab (one each). Unpaired t tests showed that expression
ofPD-L1 was significantly higher in patients who achieved
anobjective response (complete or partial) than in
nonrespondingpatients (Fig. 3). PD-L1 expression in the stromal
compartment,however, better discriminated between responders and
nonre-sponders than expression in tumor cells, with either the 5H1
orE1L3N antibodies.
We next assessed the association between continuous PD-L1scores
and PFS or OS by Cox univariate analysis. High PD-L1expression was
associated with improved PFS and OS and thisresult was consistent
between the two antibodies, particularlywhen PD-L1 was analyzed in
the stromal compartment (Table 1).Kaplan–Meier survival curves for
PD-L1 scores dichotomized bythe median and PFS or OS are shown in
Figures 4 and 5. Witheither antibody, PD-L1 expression in nontumor
cells was morestrongly associated with outcome than expression in
the S100positive cells. In addition, we note that the percent area
of PD-L1-expressing cells in this cohort ranged from 0% to 27.88%
(medi-an-1.93%,mean-4.72%). Tumors with PD-L1 intensity above
themedian had a corresponding mean PD-L1þ percent area of
8.47compared to 0.44 in low expressers.
On multivariable analysis, which included age, gender, LDHand
American Joint Committee on Cancer (AJCC)M stage, PD-L1expression
was an independent predictor of prolonged PFS or OS(Supplementary
Tables S2 and S3, respectively). All other vari-ables included in
the model were not associated with PFS or OS.To assess the
association between PD-L1 expression and otherclinical parameters,
we used the Student t-test. There was nosignificant association
between PD-L1 and age, gender, LDH orAJCC M stage (data not
shown).
NSCLC
H82
7H
CC
2935
H20
30H
358
H52
0H
596
PC9
H46
0SW
900
HC
C22
79M
EL62
4(+)
MEL
624(
-)76
9-P
786-
OC
AK
I-2C
AK
I-1A
498
AC
HN
YUH
OIN
YUM
AC
YUM
UL
YUM
UT
YUSA
CYU
SIT
YUR
IFYU
GA
SP
E1L3
N A
ntib
ody
5H1
Ant
ibod
y
RCC Melanoma
b-Actin
b-Actin
PD-L1
PD-L1
Figure 2.
PD-L1 levels in a panel of tumor-derivedmetastaticmelanoma,
NSCLC,or renal cell carcinoma cell lines.MEL-624 cell lines,
overexpressing ornot overexpressing PD-L1 were usedas controls
(boxed). Protein wasextracted from NSCLC, renal cellcarcinoma, and
melanoma cell linesand subjected to SDS-PAGE andWestern blot
analysis to detect PD-L1with either the 5H1 or E1L3N antibody.The
level of PD-L1 normalized to theb-actin level in each sample
wasmeasured by ImageJ software.
Kluger et al.
Clin Cancer Res; 23(15) August 1, 2017 Clinical Cancer
Research4274
on March 29, 2021. © 2017 American Association for Cancer
Research. clincancerres.aacrjournals.org Downloaded from
Published OnlineFirst February 21, 2017; DOI:
10.1158/1078-0432.CCR-16-3146
http://clincancerres.aacrjournals.org/
-
Quantification and characterization of TILs and associationwith
PD-L1 expression
To characterize T cell infiltrates in the pretreatment
specimensand assess correlations with response to therapy, the TMAs
werestained with antibodies to CD3, CD4, CD8 and Foxp3. Stainingwas
membranous for CD3, CD8 and CD4 while FOXP3 showedexclusively
nuclear staining. TIL subpopulations were character-ized by the
percent area of each, CD3, CD4, CD8 and Foxp3expressing cells
within the total area of the histospot, as done inprevious studies
(28). Our group and others have previouslyshown that PDL1
expression correlates with the overall T cellinfiltrate (29). In
this cohort we further confirmed the associationbetween high PD-L1
expression in tumor cells or stroma and highTIL densities as
measured by CD3 positivity (SupplementaryFigure S4). TIL
populations were split in two groups with highand low TIL content
determined by the median % area value. Noassociation was found
between the density of TIL subtypes or TILratios and response to
therapy, PFS or OS.
DiscussionPD-L1 expression is being used as a predictive
biomarker for
immune checkpoint inhibitors or incorporated into
multipara-metric predictive biomarker models in numerous diseases.
Herewe show that expression of PD-L1 varies across tumor types.
Theexpression was globally markedly higher in NSCLC tumors
thanrenal cell carcinoma tumors, and expression was lower in
melanomas. Moreover, the pattern of expression appears to
differbetween the three tumor types, with more expression on
non-tumor cells within the tumor microenvironment in
melanomasamples than in NSCLC or renal cell carcinoma. Most
clinicallyused assays assess the overall percent of positively
staining cellswithin the entire tumor to develop predictive
biomarker models,and we therefore compared expression of PD-L1 in
the tumormicroenvironment and in tumor cells in pretreatment
samplesfrompatients treatedwith immune checkpoint inhibitors.
Expres-sion in the tumor microenvironment appears to better
predictresponse to therapy.
The confusion related to PD-L1 as a biomarker is multifac-eted,
and due to technical issues of location and size of biopsyspecimen,
antibody variability and interpretation of results.Clinical trials
using PD-L1 or PD-1 inhibitors have used avariety cut-points to
determine "positive" versus "negative"expression, ranging from as
low as 1% in the phase III trialof nivolumab versus everolimus in
renal cell carcinoma, inwhich 75% of patients had
-
domain of PD-L1 (5H1) and the other than binds to
theintracellular domain, E1L3N. Both antibodies were shown tobe
highly specific for the target, and staining patterns betweenthe
two did correlate in general. When comparing the threetumor types
for differences in global PD-L1 expression usinguniform conditions
with either antibody, we found that PD-L1expression was
significantly higher in NSCLC than renal cellcarcinoma or melanoma,
and somewhat higher in renal cellcarcinoma than melanoma. These
differences might account for
the higher cut-points used in some NSCLC trials compared
tomelanoma, as our data indicate that a cut-point of 50%
inmelanoma, would render all patients "negative" for
PD-L1expression.
Expressionof PD-L1was stronger in apanel ofNSCLCcell
lines,intermediate in renal cell carcinoma cell lines,
whereasmelanomacell lines were least likely to express PD-L1, and
PD-L1þ cell lineshad lower expression intensity than NSCLC or renal
cell carcino-ma, further indicating that activity of PD-1 or PD-L1
inhibitors in
Figure 4.
Kaplan–Meier curves showing theassociation between
PD-L1expression and PFS. The medianPD-L1 intensity score,measured
eitherin tumor cells or stroma by the 5H1 orE1L3N antibodies, was
utilized todichotomize our cohort into low/highcategories. Patients
with high PD-L1levels in the stromal compartment hada longer median
PFS compared to lowexpressers.
Figure 5.
Kaplan–Meier curves showing theassociation between PD-L1
expressionand OS. PD-L1 scores weredichotomized by the median score
inthe tumor or stromal compartment.High PD-L1 level in the
stromalcompartment was significantlyassociated with prolonged
OS.
Kluger et al.
Clin Cancer Res; 23(15) August 1, 2017 Clinical Cancer
Research4276
on March 29, 2021. © 2017 American Association for Cancer
Research. clincancerres.aacrjournals.org Downloaded from
Published OnlineFirst February 21, 2017; DOI:
10.1158/1078-0432.CCR-16-3146
http://clincancerres.aacrjournals.org/
-
melanoma might be due to the interaction between PD-L1
innontumor cells and PD-1 in TILs.
Our second purpose was to study the clinical significance
ofPD-L1 expression in the tumor versus the tumor microenvi-ronment
in melanoma samples. We used TMAs to determinePD-L1 expression in
up to three locations in the tumor, as thediameter of TMA
histocores is similar to that of biopsies,rendering them a useful
surrogate for clinical evaluation ofmetastatic samples, which is
often based on biopsy rather thanmetastatectomy. Our studies were
therefore not designed todetermine location of PD-L1þ cells within
the tumor bed, andcannot be compared to findings of Taube and
colleagues whohave shown that an abundance of PD-L1þ cells at the
tumoredge, rather than deep with the tumor, is associated
withresponse to therapy (35, 37).
We examined PD-L1 expression in either melanoma cells ornontumor
cells separately for each antibody. Our studiesincluded only 34
patients who received anti-PD-1 monother-apy, of whom only 20 had
not received ipilimumab within 3months prior to initiation of PD-1
inhibitors. In this smallgroup of 20, PD-L1 expression was weakly
associated withresponse to therapy, with a P value that only
trended towardsignificance. Given the small sample size and the
relatively lowresponse rate to therapy (
-
Grant SupportThis work was supported in part by NIH grantsR0-1
CA158167, NIH
R01CA216846, and K24CA172123 (to H. Kluger, PI), Yale SPORE in
SkinCancer, P50 CA121974 (to R. Halaban, PI), NIH R01CA204002, CTSA
GrantNumber KL2 TR000140 (to L. Jilaveanu, PI) from the National
Center forResearch Resources (NCRR) and the National Center for
Advancing Transla-tional Science (NCATS), components of theNational
Institutes ofHealth (NIH),and NIH roadmap for Medical Research, a
grant from the Lung Cancer research
Foundation-LUNGevity and Melanoma Research Alliance, Award
#308721 (toL. Jilaveanu, PI), and in part by a Research Scholar
Grant (130157-RSG-16-216-01-TBG) from the American Cancer Society
(to L. Jilaveanu, PI). The content ofthis publication is solely the
responsibility of the authors and does notnecessarily represent the
official views of the National Institutes of Health.
Received December 14, 2016; revised January 5, 2017; accepted
February 17,2017; published OnlineFirst February 21, 2017.
References1. Larkin J, Chiarion-Sileni V, Gonzalez R, Grob JJ,
Cowey CL, Lao CD, et al.
Combined Nivolumab and Ipilimumab or Monotherapy in
UntreatedMelanoma. N Engl J Med 2015;373:23–34.
2. Hodi FS,O'Day SJ,McDermott DF,Weber RW, Sosman JA,Haanen JB,
et al.Improved survivalwith ipilimumab in patients
withmetastaticmelanoma.N Engl J Med 2010;363:711–23.
3. Robert C, Thomas L, Bondarenko I, O'Day S, Weber J, Garbe C,
et al.Ipilimumab plus dacarbazine for previously untreated
metastatic mela-noma. N Engl J Med 2011;364:2517–26.
4. Hamid O, Robert C, Daud A, Hodi FS, HwuWJ, Kefford R, et al.
Safety andtumor responses with lambrolizumab (anti-PD-1) in
melanoma. N Engl JMed 2013;369:134–44.
5. Robert C, Long GV, Brady B, Dutriaux C, Maio M, Mortier L, et
al.Nivolumab in previously untreated melanoma without BRAF
mutation.N Engl J Med 2015;372:320–30.
6. Weber JS, D'Angelo SP, Minor D, Hodi FS, Gutzmer R, Neyns B,
et al.Nivolumab versus chemotherapy in patients with advanced
melanomawho progressed after anti-CTLA-4 treatment (CheckMate 037):
a rando-mised, controlled, open-label, phase 3 trial. Lancet Oncol
2015;16:375–84.
7. Motzer RJ, Escudier B, McDermott DF, George S, Hammers HJ,
Srinivas S,et al. Nivolumab versus everolimus in advanced
renal-cell carcinoma. NEngl J Med 2015;373:1803–13.
8. Rosenberg JE, Hoffman-Censits J, Powles T, van der Heijden
MS, Balar AV,Necchi A, et al. Atezolizumab in patients with locally
advanced andmetastatic urothelial carcinoma who have progressed
following treatmentwith platinum-based chemotherapy: a single-arm,
multicentre, phase 2trial. Lancet 2016;387:1909–20.
9. Yared JA, Hardy N, Singh Z, Hajj S, Badros AZ, Kocoglu M, et
al. Majorclinical response to nivolumab in relapsed/refractory
Hodgkin lymphomaafter allogeneic stem cell transplantation. Bone
Marrow Transplant 2016;51:850–2.
10. Borghaei H, Paz-Ares L, Horn L, Spigel DR, Steins M, Ready
NE, et al.Nivolumab versus docetaxel in advanced nonsquamous
non-small-celllung cancer. N Engl J Med 2015;373:1627–39.
11. Garon EB, Rizvi NA, Hui R, Leighl N, Balmanoukian AS, Eder
JP, et al.Pembrolizumab for the treatment of non-small-cell lung
cancer. N Engl JMed 2015;372:2018–28.
12. Ferris RL, Blumenschein G Jr., Fayette J, Guigay J, Colevas
AD, Licitra L,et al. Nivolumab for recurrent squamous-cell
carcinoma of the headand neck. N Engl J Med 2016;375:1856–67.
13. PostowMA, Chesney J, Pavlick AC, Robert C, Grossmann
K,McDermott D,et al. Nivolumab and ipilimumab versus ipilimumab in
untreated mela-noma. N Engl J Med 2015;372:2006–17.
14. Wolchok JD, Kluger H, CallahanMK, PostowMA, Rizvi NA,
Lesokhin AM,et al. Nivolumab plus ipilimumab in advanced melanoma.
N Engl J Med2013;369:122–33.
15. Yuan J, Adamow M, Ginsberg BA, Rasalan TS, Ritter E,
Gallardo HF, et al.Integrated NY-ESO-1 antibody and CD8þ T-cell
responses correlate withclinical benefit in advanced melanoma
patients treated with ipilimumab.Proc Natl Acad Sci U S A
2011;108:16723–8.
16. Martens A, Wistuba-Hamprecht K, Geukes Foppen M, Yuan J,
Postow MA,Wong P, et al. Baseline peripheral blood biomarkers
associated withclinical outcomeof advancedmelanomapatients
treatedwith ipilimumab.Clin Cancer Res 2016;22:2908–18.
17. Ribas A, Robert C, Hodi FS, Wolchok JD, Joshua AM, Hwu W-J,
et al.Association of response to programmed death receptor 1 (PD-1)
blockadewith pembrolizumab (MK-3475) with an
interferon-inflammatory
immune gene signature. ASCO Annual Meeting; J Clin Oncol
2015;33:suppl; abstr 3001.
18. Hugo W, Zaretsky JM, Sun L, Song C, Moreno BH, Hu-Lieskovan
S, et al.Genomic and transcriptomic features of response to
anti-PD-1 therapy inmetastatic melanoma. Cell 2016;165:35–44.
19. Tumeh PC, Harview CL, Yearley JH, Shintaku IP, Taylor EJ,
Robert L, et al.PD-1 blockade induces responses by inhibiting
adaptive immune resis-tance. Nature 2014;515:568–71.
20. Johnson DB, Estrada MV, Salgado R, Sanchez V, Doxie DB,
Opalenik SR,et al. Melanoma-specific MHC-II expression represents a
tumour-autono-mous phenotype and predicts response to
anti-PD-1/PD-L1 therapy. NatCommun 2016;7:10582.
21. Wang X, Teng F, Kong L, Yu J. PD-L1 expression in human
cancers andits association with clinical outcomes. Onco Targets
Ther 2016;9:5023–39.
22. Festino L, Botti G, Lorigan P, Masucci GV, Hipp JD, Horak
CE, et al. CancerTreatment with Anti-PD-1/PD-L1 Agents: Is PD-L1
expression a biomarkerfor patient selection? Drugs
2016;76:925–45.
23. Topalian SL, Taube JM, Anders RA, Pardoll DM.
Mechanism-driven bio-markers to guide immune checkpoint blockade in
cancer therapy. Nat RevCancer 2016;16:275–87.
24. McLaughlin J, Han G, Schalper KA, Carvajal-Hausdorf D,
Pelekanou V,Rehman J, et al. Quantitative assessment of the
heterogeneity of PD-L1expression in non-small-cell lung cancer.
JAMA Oncol 2016;2:46–54.
25. TopalianSL,Hodi FS, Brahmer JR,Gettinger SN,
SmithDC,McDermottDF,et al. Safety, activity, and immune correlates
of anti-PD-1 antibody incancer. N Engl J Med 2012;366:2443–54.
26. Herbst RS, Soria JC, Kowanetz M, Fine GD, Hamid O, Gordon
MS, et al.Predictive correlates of response to the anti-PD-L1
antibody MPDL3280Ain cancer patients. Nature 2014;515:563–7.
27. Jilaveanu LB, Shuch B, Zito CR, Parisi F, Barr M, Kluger Y,
et al. PD-L1expression in clear cell renal cell carcinoma: an
analysis of nephrectomyand sites of metastases. J Cancer
2014;5:166–72.
28. Baine MK, Turcu G, Zito CR, Adeniran AJ, Camp RL, Chen L, et
al.Characterization of tumor infiltrating lymphocytes in paired
primaryand metastatic renal cell carcinoma specimens. Oncotarget
2015;6:24990–5002.
29. Kluger HM, Zito CR, Barr ML, Baine MK, Chiang VL, Sznol M,
et al.Characterization of PD-L1 expression and associated T-cell
infiltrates inmetastatic melanoma samples from variable anatomic
sites. Clin CancerRes 2015;21:3052–60.
30. Jilaveanu LB, Zito CR, Aziz SA, Conrad PJ, Schmitz JC, Sznol
M, et al. C-Rafis associated with disease progression and cell
proliferation in a subset ofmelanomas. Clin Cancer Res
2009;15:5704–13.
31. McCarthy MM, DiVito KA, Sznol M, Kovacs D, Halaban R, Berger
AJ, et al.Expression of tumor necrosis factor–related
apoptosis-inducing ligandreceptors 1 and 2 in melanoma. Clin Cancer
Res 2006;12:3856–63.
32. McCarthy MM, Pick E, Kluger Y, Gould-Rothberg B, Lazova R,
Camp RL,et al. HSP90 as a marker of progression in melanoma. Ann
Oncol2008;19:590–4.
33. Camp RL, Chung GG, Rimm DL. Automated subcellular
localization andquantification of protein expression in tissue
microarrays. Nat Med2002;8:1323–7.
34. Dong H, Strome SE, Salomao DR, Tamura H, Hirano F, Flies DB,
et al.Tumor-associated B7-H1 promotes T-cell apoptosis: a potential
mecha-nism of immune evasion. Nat Med 2002;8:793–800.
35. Taube JM, Klein A, Brahmer JR, Xu H, Pan X, Kim JH, et al.
Association ofPD-1, PD-1 ligands, and other features of the tumor
immune
Clin Cancer Res; 23(15) August 1, 2017 Clinical Cancer
Research4278
Kluger et al.
on March 29, 2021. © 2017 American Association for Cancer
Research. clincancerres.aacrjournals.org Downloaded from
Published OnlineFirst February 21, 2017; DOI:
10.1158/1078-0432.CCR-16-3146
http://clincancerres.aacrjournals.org/
-
microenvironment with response to anti-PD-1 therapy. Clin Cancer
Res2014;20:5064–74.
36. Mahoney KM, Sun H, Liao X, Hua P, Callea M, Greenfield EA,
et al. PD-L1antibodies to its cytoplasmic domain most clearly
delineate cell mem-branes in immunohistochemical staining of tumor
cells. Cancer ImmunolRes 2015;3:1308–15.
37. Taube JM, Anders RA, Young GD, Xu H, Sharma R, McMiller TL,
et al.Colocalization of inflammatory response with B7-h1 expression
in humanmelanocytic lesions supports an adaptive resistance
mechanism ofimmune escape. Sci Transl Med 2012;4:127ra37.
38. Daud AI, Wolchok JD, Robert C, Hwu WJ, Weber JS, Ribas A, et
al.Programmed death-ligand 1 expression and response to the
anti-programmed death 1 antibody pembrolizumab in melanoma.J
Clin Oncol 2016;34:4102–9.
39. Dong H, Zhu G, Tamada K, Chen L. B7-H1, a third member of
the B7family, co-stimulates T-cell proliferation and interleukin-10
secretion. NatMed 1999;5:1365–9.
40. Parsa AT, Waldron JS, Panner A, Crane CA, Parney IF, Barry
JJ, et al. Loss oftumor suppressor PTEN function increases B7-H1
expression and immu-noresistance in glioma. Nat Med
2007;13:84–8.
41. Gowrishankar K, Gunatilake D, Gallagher SJ, Tiffen J, Rizos
H, Hersey P.Inducible but not constitutive expression of PD-L1 in
human melanomacells is dependent on activation of NF-kappaB. PLoS
One 2015;10:e0123410.
www.aacrjournals.org Clin Cancer Res; 23(15) August 1, 2017
4279
PD-L1 in the Tumor Microenvironment
on March 29, 2021. © 2017 American Association for Cancer
Research. clincancerres.aacrjournals.org Downloaded from
Published OnlineFirst February 21, 2017; DOI:
10.1158/1078-0432.CCR-16-3146
http://clincancerres.aacrjournals.org/
-
2017;23:4270-4279. Published OnlineFirst February 21, 2017.Clin
Cancer Res Harriet M. Kluger, Christopher R. Zito, Gabriela Turcu,
et al. InhibitorsPredictive Value in Patients Treated with Immune
Checkpoint PD-L1 Studies Across Tumor Types, Its Differential
Expression and
Updated version
10.1158/1078-0432.CCR-16-3146doi:
Access the most recent version of this article at:
Material
Supplementary
http://clincancerres.aacrjournals.org/content/suppl/2017/02/21/1078-0432.CCR-16-3146.DC1
http://clincancerres.aacrjournals.org/content/suppl/2017/07/04/1078-0432.CCR-16-3146.DC2
Access the most recent supplemental material at:
Cited articles
http://clincancerres.aacrjournals.org/content/23/15/4270.full#ref-list-1
This article cites 41 articles, 8 of which you can access for
free at:
Citing articles
http://clincancerres.aacrjournals.org/content/23/15/4270.full#related-urls
This article has been cited by 11 HighWire-hosted articles.
Access the articles at:
E-mail alerts related to this article or journal.Sign up to
receive free email-alerts
Subscriptions
Reprints and
[email protected]
To order reprints of this article or to subscribe to the
journal, contact the AACR Publications Department at
Permissions
Rightslink site. Click on "Request Permissions" which will take
you to the Copyright Clearance Center's (CCC)
.http://clincancerres.aacrjournals.org/content/23/15/4270To
request permission to re-use all or part of this article, use this
link
on March 29, 2021. © 2017 American Association for Cancer
Research. clincancerres.aacrjournals.org Downloaded from
Published OnlineFirst February 21, 2017; DOI:
10.1158/1078-0432.CCR-16-3146
http://clincancerres.aacrjournals.org/lookup/doi/10.1158/1078-0432.CCR-16-3146http://clincancerres.aacrjournals.org/content/suppl/2017/07/04/1078-0432.CCR-16-3146.DC2http://clincancerres.aacrjournals.org/content/suppl/2017/02/21/1078-0432.CCR-16-3146.DC1http://clincancerres.aacrjournals.org/content/23/15/4270.full#ref-list-1http://clincancerres.aacrjournals.org/content/23/15/4270.full#related-urlshttp://clincancerres.aacrjournals.org/cgi/alertsmailto:[email protected]://clincancerres.aacrjournals.org/content/23/15/4270http://clincancerres.aacrjournals.org/
/ColorImageDict > /JPEG2000ColorACSImageDict >
/JPEG2000ColorImageDict > /AntiAliasGrayImages false
/CropGrayImages false /GrayImageMinResolution 200
/GrayImageMinResolutionPolicy /Warning /DownsampleGrayImages true
/GrayImageDownsampleType /Bicubic /GrayImageResolution 300
/GrayImageDepth -1 /GrayImageMinDownsampleDepth 2
/GrayImageDownsampleThreshold 1.50000 /EncodeGrayImages true
/GrayImageFilter /DCTEncode /AutoFilterGrayImages true
/GrayImageAutoFilterStrategy /JPEG /GrayACSImageDict >
/GrayImageDict > /JPEG2000GrayACSImageDict >
/JPEG2000GrayImageDict > /AntiAliasMonoImages false
/CropMonoImages false /MonoImageMinResolution 600
/MonoImageMinResolutionPolicy /Warning /DownsampleMonoImages true
/MonoImageDownsampleType /Bicubic /MonoImageResolution 900
/MonoImageDepth -1 /MonoImageDownsampleThreshold 1.50000
/EncodeMonoImages true /MonoImageFilter /CCITTFaxEncode
/MonoImageDict > /AllowPSXObjects false /CheckCompliance [ /None
] /PDFX1aCheck false /PDFX3Check false /PDFXCompliantPDFOnly false
/PDFXNoTrimBoxError true /PDFXTrimBoxToMediaBoxOffset [ 0.00000
0.00000 0.00000 0.00000 ] /PDFXSetBleedBoxToMediaBox true
/PDFXBleedBoxToTrimBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ]
/PDFXOutputIntentProfile (None) /PDFXOutputConditionIdentifier ()
/PDFXOutputCondition () /PDFXRegistryName () /PDFXTrapped
/False
/CreateJDFFile false /Description > /Namespace [ (Adobe)
(Common) (1.0) ] /OtherNamespaces [ > /FormElements false
/GenerateStructure false /IncludeBookmarks false /IncludeHyperlinks
false /IncludeInteractive false /IncludeLayers false
/IncludeProfiles false /MarksOffset 18 /MarksWeight 0.250000
/MultimediaHandling /UseObjectSettings /Namespace [ (Adobe)
(CreativeSuite) (2.0) ] /PDFXOutputIntentProfileSelector /NA
/PageMarksFile /RomanDefault /PreserveEditing true
/UntaggedCMYKHandling /LeaveUntagged /UntaggedRGBHandling
/LeaveUntagged /UseDocumentBleed false >> > ]>>
setdistillerparams> setpagedevice