Oral Oncology 101 (2020) 104516

Contents lists available at ScienceDirect

Oral Oncology
journal homepage: www.elsevier.com/locate/oraloncology

Validation and characterisation of prognostically significant PD-L1+ T

immune cells in HPV+ oropharyngeal squamous cell carcinoma
Richard J. Younga, Mathias Bresselb, Sandro Porcedduc,d, Janez Cernelce, Peter Savasa,f,
Howard Liuc,d, Damien Urbang, Alesha A. Thaia,f, Caroline Cooperd,h, Tsien Fuai, Paul Neesona,j,
⁎
Danny Rischinf,j, Benjamin Solomona,f,j,
a
Research Division, Peter MacCallum Cancer Centre, Melbourne, Australia
b
Centre for Biostatistics and Clinical Trials, Peter MacCallum Cancer Centre, Melbourne, Australia
c
Department of Radiation Oncology, Princess Alexandra Hospital, Brisbane, Australia
d
Faculty of Medicine, University of Queensland, Brisbane, Australia
e
Alfred Hospital, Prahran, Australia
f
Department of Medical Oncology, Peter MacCallum Cancer Centre, Melbourne, Australia
g
Department of Medical Oncology, Sheba Medical Center, Israel
h
Department of Anatomical Pathology, Pathology Queensland, Princess Alexandra Hospital, Brisbane, Australia
i
Department of Radiation Oncology, Peter MacCallum Cancer Centre, Melbourne, Australia
j
Sir Peter MacCallum Department of Oncology, University of Melbourne, Melbourne, Australia

A R T I C LE I N FO A B S T R A C T

Keywords: We previously showed in human papillomavirus positive oropharyngeal squamous cell carcinoma (HPV
PD-L1 +OPSCC) that the presence of intratumoral (IT) PD-L1+ immune cells (ICs) or CD8+ infiltrating ICs are of
CD8 prognostic value. Here we report the prognostic significance of these immune biomarkers in an independent
Human papillomavirus validation cohort of 177 HPV+OPSCC patients. IT and stromal (S) localisation of PD-L1+ and CD8+ ICs were
Immune cells
scored. High abundance (≥5%) of PD-L1+ IT ICs was found in 51/167 patients (30.5%) and was associated with
Oropharynx
Survival
improved overall survival (OS) (HR, 0.21; 95% CI, 0.05–0.91; P = 0. 012) validating our previous results. High
abundance (≥30%) of CD8+ IT or S ICs, found in 77/167 patients (46.1%) provided a HR of 0.45 for OS
however the confidence interval was wide (95% CI 0.16–1.25, p = 0.105). Multiplex immunohistochemistry
revealed CD68+ macrophages and CD3+CD8+ T cells to be the most common ICs expressing PD-L1. Gene
expression analysis showed tumors with high abundance of PD-L1+ IT ICs exhibit gene signatures associated
with responses to PD1 or PD-L1 inhibitors pembrolizumab and atezolizumab. These data support the role of
immune biomarkers such as PD-L1+ ICs to identify subgroups of HPV+OPSCC patients with an excellent out-
come that may be suitable for trials evaluating de-intensification of therapy.

Introduction interest in de-intensified treatment approaches for HPV+OPSCC pa-

Human papillomavirus (HPV) is responsible for an increasing pro-
portion of head and neck squamous cell carcinomas and accounts for
the majority of oropharyngeal cancers in many western countries in-
cluding the United States [1,2]. HPV positive (HPV+) tumors arise in
the tonsil and base of tongue, respond better to conventional treatments
than HPV negative (HPV−) tumors and importantly, are associated
with improved survival compared with HPV− tumors [3]. Given the
improved outcomes for HPV+OPSCC patients and the significant acute
and late toxicities associated with conventional treatments, such as
radiotherapy with or without chemotherapy, there is considerable
tients. Biomarker determined subgroups may better define a low risk
populations suitable for de-intensification trials, compared to currently
available TNM-based staging criteria [4,5].
We previously showed in a cohort of 190 patients with HPV
+OPSCC that the presence of PD-L1+ immune cells (ICs) found spe-
cifically within the intratumoral (IT) nests is strongly prognostic for
improved survival [6]. We also demonstrated that CD8+ IC’s found in
the IT nests or surrounding stroma was similarly associated with im-
proved survival. Here we present further work, in an independent co-
hort of HPV+OPSCC patients, that confirms and validates our previous
findings. Further, we have used multiplex immunohistochemistry and

⁎
Corresponding author at: Peter MacCallum Cancer Centre, 305 Grattan Street, Melbourne, Victoria 3000, Australia.
E-mail address: ben.solomon@petermac.org (B. Solomon).

https://doi.org/10.1016/j.oraloncology.2019.104516
Received 3 December 2019; Accepted 4 December 2019
Available online 12 December 2019
1368-8375/ © 2019 Published by Elsevier Ltd.
R.J. Young, et al.
gene expression profiling to characterise the immune cell phenotypes of
the PD-L1+ ICs.
Materials and methods
Patients
We previously reported findings from a retrospective cohort of 190
HPV+OPSCC patients planned for curative treatment at the Peter
MacCallum Cancer Centre between 2002 and 212 (PMCC cohort) [6].
The validation cohort for the present study comprised an independent
cohort of 177 patients from Princess Alexandra Hospital treated with
definitive radiotherapy +/- chemotherapy between 2008 and 2015,
with prospectively collected clinical outcome data (PAH cohort). Tumor
blocks (formalin-fixed paraffin embedded) or unstained tissue sections,
together with clinico-pathologic and outcome data, were collected. p16
immunohistochemistry (IHC) was performed to confirm HPV status. All
samples included in the study were from primary resections or biopsies
taken prior to the commencement of treatment. The study had In-
stitutional Ethics Committee approval (PMCC HREC 12/144).
Immunohistochemistry
Immunohistochemistry (IHC) for p16, CD8 and PD-L1 was per-
formed as previously described [6] on 4 μm whole tumor sections.
Slides were baked at 60 °C for 1 h, followed by dewaxing through
several xylene baths, graded alcohols and water. Antigen retrieval was
performed in a pressure cooker at 125 °C for 3 min after which slides
were placed onto a Dako autostainer for the following incubations: 3%
H2O2 for 10 min, primary antibody: p16 - Cintec Histology kit (Ven-
tana) for 60 min; PD-L1 - (1:500; clone SP142, Abcam) for 10 min; CD8
− 1:1000; clone 4B11, Novocastra) for 60 min, secondary detection:
p16 - Cintec Histology kit for 60 min; PD-L1 - Envision + rabbit (Dako)
for 30 min; CD8 - Envision + mouse (Dako) for 60 min, colour detec-
tion with 3,3′-diaminobenzidine (DAB) for 10 min. Slides were rinsed
with TBST between each incubation. After IHC, all slides were coun-
terstained with haemotoxylin, mounted and coverslipped.
Immunohistochemical scoring
p16 expression was scored for both intensity and proportion of
staining in the cell nucleus and cytoplasm. Intensity was scored as 0
(none), 1 (weak), 2 (moderate), or 3 (strong); with proportion scored as
0–100% to the nearest 10% for each intensity. The prevalence and tu-
moral location of PD-L1+ and CD8+ ICs were determined by semi-
quantitatively scoring the proportion of positively stained ICs located
(a) within the actual tumor nests (intratumoral) and (b) in the stroma
immediately adjacent to the tumor nests. PD-L1 was scored for the
proportion of positively stained immune cells (ICs) as 0 (< 1%), 1
(1–4%), 2 (5–10%) or 3 (> 10%) [7]. The expression of PD-L1 on tumor
cells was also scored as 0, 1, 5, 10, 20, 30… −100% (to the nearest
10%). The proportion of positively stained CD8+ ICs was scored as 0, 1,
5, 10, 20, 30… −100% (to the nearest 10%). All IHC was scored by two
independent observers; any discrepant results were reviewed together
by both observers and a consensus score was obtained.
Multiplex immunohistochemistry
OPAL multiplex immunohistochemistry (mIHC) was used to enable
simultaneous detection, imaging, and analysis, on single sections, of
PD-L1 expression by immune cells. Specific immune cell markers used
were CD3 (T cells), CD8 (cytotoxic T cells), CD4 (T helper cells), CD68
(macrophages), CD19 (B cells) and CD11c (dendritic cells). p16 (for
identification of tumor nests) and DAPI (for cell visualization) were also
included. All mIHC was performed manually or on a BOND autostainer
(Leica). Slides were baked prior to antigen retrieval followed by the first
2
Oral Oncology 101 (2020) 104516
primary antibody for 30 min; 0.1% H2O2 block for 5 min; secondary
antibody (rabbit or mouse horseradish peroxidase–conjugated sec-
ondary antibodies at a 1:500 dilution) for 10 min; TSA reagent
(PerkinElmer; diluted 1:150 in TSA amplification diluent) for 10 min.
Three 2-minute washes in TBS were performed between each step
above. At the completion of the above process, microwave heat removal
of 1° and 2° antibodies was performed, which, after cooling, enabled the
protocol to be repeated for multiplexing of subsequent primary anti-
bodies. Primary antibodies used were p16 (predilute; CINtec); PD-L1
(1:500; Abcam); CD3 (1:500; Abcam), CD4 (1:100; Spring Bioscience),
CD8 (1:1000; Leica Biosystems), CD68 (1:200; Leica Biosystems), CD19
(1:200; Dako) and CD11c (1:2000; Bio SB). After the final antibody
protocol was completed, slides were stained with DAPI for one minute,
mounted, coverslipped and stored at 4 °C in the dark. Single antibody
control slides were included for each antibody individually, and were
used as imaging controls.
Immune cell phenotyping
A Vectra microscope (Olympus) utilizing Vectra 3.0.3 software was
used to image mIHC slides. A low-power overview image of the entire
section was obtained at ×4 magnification, which was used to guide
selection of ×20 magnification high power multispectral images
(MSIs). Inform software (V2.04.01) was used to spectrally unmix the
fluorescent signals, false color the images and create MSIs for cell
phenotype analysis. HALO software (Indica Labs) was used to semi-
quantitatively analyse the co-localisation of PD-L1 and various immune
cell type markers, in order to determine which immune cells were ex-
pressing PD-L1. Briefly, for each slide, up to six different fluorescent
markers (including DAPI) were imaged and combined into a single
fused image for analysis. HALO software was trained to identify and
classify tissue areas as tumor or stroma, followed by identification of
specific cell phenotypes based on fluorescence intensity and positivity
to the various fluorescent markers. For example, using dapi, p16, CD3,
CD8, CD68 and PD-L1, individual cells located in the intra-tumoral
compartment could be phenotyped as being tumor cells (p16+ or
p16+ PD-L1+), T cells (CD3+ or CD3+PD-L1+), cytotoxic T cells
(CD3+CD8+ or CD3+CD8+PD-L1+ or macrophages (CD68+ or
CD68+PD-L1+).
Gene expression analysis
RNA was extracted from tumors from the PMCC cohort using the
Roche High Pure FFPE RNA micro kit (Roche). Extracted RNA was re-
presentative of tumor cells and other cells of the tumor microenviron-
ment. Gene expression was analysed for 74 HPV+OPSCC samples using
the Nanostring Gene Expression Platform and the nCounter Human
PanCancer Immune Panel, which comprises 730 immune related genes
and 40 control genes. Raw data was normalised using nSolver software
(Nanostring). Differential expression analysis of the normalised data
was performed using the nSolver Advanced Analysis softwares. P-values
were adjusted according to the method of Benjamini-Yekutieli in
nSolver. Volcano plots of log fold change versus log10 unadjusted p-
value were made using nSolver differential expression output and the R
packages ‘ggplot2’ v3.0 and ‘ggrepel’ v0.8.0. For visualization purposes,
gene-wise z-scores were calculated prior to plotting. The R package
‘superheat’ version 0.1.0 was used for producing heatmaps. The two-
sided Fisher’s exact test was used to compare PD-L1 high and low cases.
For gene set testing and barcode plot, methods available in the R
package ‘limma’ v3.36.3 were used. Raw counts were TMM normalised
followed by transformation with voom prior to gene set testing with
ROAST.
Statistics
Both PD-L1+ and CD8+ raw data were dichotomized into low or
R.J. Young, et al. Oral Oncology 101 (2020) 104516

Table 1
Patient characteristics of the PAH and PMCC cohorts. *PMCC cohort previously
described [6].
Variable PAH (n = 177) PMCC* (n = 189)

Age
Mean (SD) 59 (8.7) 57 (9.2)
Median [range] 59 [21–89] 57 [34–85]
Interquartile range 53–65 52–62

Gender
Female 20 (11.3%) 33 (17.5%)
Male 157 (88.7%) 156 (82.5%)

Smoker
No 44 (24.9%) 67 (35.4%)
Yes 133 (75.1%) 122 (64.6%)

Smoker > 10 Packs/Year

No 92 (52.0%) 104 (56.8%)
Yes 85 (48.0%) 79 (43.2%)

Chemotherapy
No 9 (5.1%) 26 (13.8%)
Yes 168 (94.9%) 163 (86.2%)

AJCC 7th stage

1 0 (0%) 1 (0.5%)
2 4 (2.3%) 10 (5.3%)
3 19 (10.7%) 25 (13.2%)
4 154 (87.0%) 153 (81.0%)

AJCC 8th stage

1 87 (49.2%) 93 (49.2%)
2 53 (29.9%) 66 (34.9%)
3 37 (20.9%) 30 (15.9%)

high expression, using the cut-offs defined in the previous study [6]. For
PD-L1+ ICs high abundance was defined as ≥5% and for CD8+ ICs
high abundance was defined as ≥30%. The Kaplan–Meier method was
used to describe the survival curves. OS was defined from the time of
initial diagnosis to the date of death from any cause. Cox proportional
hazard models were used to assess the impact of PD-L1+ and CD8+ on
OS. Multivariable analysis was carried out adjusting for AJCC 8th edi-
tion stage and age and likelihood ratio p-values were provided.

Results

Patient characteristics

The Peter MacCallum Cancer Centre (PMCC) cohort that comprised

190 patients with HPV+OPSCC, has been previously described [6]. The
validation cohort for the present study comprised 177 HPV+OPSCC
patients from the Princess Alexandra Hospital (PAH cohort) treated
with radiation +/- chemotherapy with curative intent with available
tissue and clinical outcome data. The clinicopathological characteristics
of the PAH cohort patients are detailed in Table 1. The median age at
diagnosis was 59 (range 21–89), 88.7% of the patients were male,
75.1% were smokers. Immunohistochemistry for p16 was able to be
performed on 167/177 patients, all 167 were confirmed p16 positive.

Prognostic significance of PD-L1+ immune cells

We previously showed that high abundance of PD-L1+ IC in the IT

compartment, found in 70/181 (38.7%) patients, was associated with
excellent OS (HR, 0.37; 95% CI, 0.15–0.93, p = 0.023) (Fig. 1A) [6].
Similar abundance of PD-L1+ IT ICs was seen in the PAH validation
cohort with 51/167 (30.5%) patients showing high abundance (defined
as ≥5%) of IT PD-L1+ IC, which was again associated with improved
OS (HR 0.21; 95% CI 0.05–0.91, p = 0.012) adjusting for stage and age
(Fig. 1B). Of the 51 patients with high abundance of PD-L1 IT ICs only
two events were recorded, compared with 19 events in the 116 patients
with low PD-L1+ IT ICs. An example of PD-L1 IHC in the PAH cohort
Fig. 1. Prognostic significance of PD-L1+ immune cells. (A) Kaplan-Meier plot
for OS according to the presence of intratumoral PD-L1+ immune cells in the
PMCC cohort. (B) Kaplan-Meier plot for OS according to the presence of in-
tratumoral PD-L1+ immune cells in the PAH cohort. For (A) and (B) red line is
high abundance of PD-L1+ ICs (≥5%); black line is low abundance (< 5%). (C)
An example of PD-L1 IHC in the PAH cohort showing high abundance of PD-
L1+ intratumoral immune cells.

3
R.J. Young, et al.
Fig. 2. Prognostic significance of CD8+ immune cells. (A) Kaplan-Meier plot for
OS according to the presence of intratumoral or stromal CD8+ immune cells in
the PMCC cohort. (B) Kaplan-Meier plot for OS according to the presence of
intratumoral or stromal CD8+ immune cells in the PAH cohort. For (A) and (B)
red line is high abundance of CD8+ ICs (≥30%); black line is low abundance
(< 30%). (C) An example of CD8 IHC in the PAH cohort showing high abun-
dance of both intratumoral and stromal CD8+ immune cells.
4
Oral Oncology 101 (2020) 104516
showing high abundance of PD-L1+ IT ICs is shown in Fig. 1C. Con-
sistent with our previous findings, neither the abundance of stromal PD-
L1+ IC’s, nor the expression of PD-L1 on tumor cells, correlated with OS
(data not shown).
Prognostic significance of CD8 ± immune cells
We and others have previously reported the prognostic significance
of CD8+ ICs in HPV+ HNSCC. In the PMCC cohort (6) where we found
that high abundance of CD8+ IC in the intratumoral (IT) or stromal (S)
compartments, found in 70/181 (38.7%) patients was associated with
improved OS adjusting for stage and age (HR, 0.42; 95% CI, 0.2–0.87,
p = 0.017) (Fig. 2A). For the PAH cohort we were able to assess CD8
IHC for 167/167 confirmed p16 positive cases, of which 77/167
(46.1%) were found to have high abundance of IT or S CD8+ IC’s,
consistent with the PMCC cohort. The results from the PAH cohort were
consistent with the PMCC cohort (HR 0.45; 95% CI, 0.16–1.25,
P = 0.105), although less precise (wider confidence interval) due to
fewer events in the PAH cohort (Fig. 2B). An example of CD8 IHC in the
PAH cohort showing high abundance of CD8+ ICs is shown in Fig. 2C.
PD-L1+ IC status impacts on TNM-based classification of HPV+OPSCC
Consistent with our previously published findings for the PMCC
cohort, we analysed the effect of the addition of PD-L1+ IC status to
AJCC 8th ed. staging of HPV+OPSCC. The C statistic for the Cox model
including stage and age alone was 0.75 in the PAH cohort and this
improved to 0.82 with the addition of PD-L1+ IT IC abundance. High
abundance of PD-L1+ IT ICs is able to identify patients with excellent
OS independent of stage with 3-year survival in each of AJCC 8th ed.
Stage I, II and III of 100%, 100% and 100% respectively, compared to
96% (95% CI, 86–99), 91% (95% CI, 75–97) and 76% (95% CI, 54–88)
for low abundance of PD-L1+ IT ICs (Fig. 3A–C), consistent with our
previous finding.
Characterisation of PD-L1+ immune cells
Given the prognostic significance of PD-L1+ ICs within the in-
tratumoral compartment of HPV+OPSCC we set out to further char-
acterise these cells through multiplex immunohistochemistry (mIHC)
and gene expression analysis. mIHC was performed on a subset of tu-
mors to determine which types of immune cells expressed PD-L1 within
the IT nests. Co-labelling of PD-L1 with immune cell markers such as
CD3 (T cells), CD8 (cytotoxic T cells), CD4 (T helper cells), CD68
(macrophages), CD19 (B cells) and CD11c (dendritic cells) enabled
semi-quantitative image analysis to be performed. Of the PD-L1+ IT ICs
we found that CD68+ macrophages were the most common immune
cell type (mean 31.7%; range 18.7–47.2%, n = 9), along with
CD3+CD8+ cytotoxic T cells (mean 23.6%; range 1.5–44.9% n = 18)
and CD3+CD4+ T helper cells (mean 20.8%; range 13.2–33.3%,
n = 5). A small proportion of the PD-L1+ ICs were CD19+ B cells
(mean 7.2%; range 0.2–13.3%, n = 9) and CD11c+ dendritic cells
(mean 2.0%; range 0.2–6.2%, n = 9). (Fig. 4A–C).
Gene expression analysis was performed comparing tumors with
abundant PD-L1+ IT ICs to tumors with low PD-L1+ IT IC abundance.
Differential expression analysis found that tumors with abundant PD-
L1+ IT ICs had significantly increased expression of many immune
related genes, including immune checkpoints PD1, IDO1, LAG3 and
HAVCR2 (Tim3) (Fig. 4D). PD-L1+ IT IC abundant tumors also showed
increased expression of genes associated with two previously described
gene signatures; a three gene ‘Teff’ signature (IFNG, CXCL9, CD274)
associated with responses to atezolizumab [8] (Fig. 4E) and a 16 gene
IFNγ related ‘T-cell inflamed’ signature predictive of clinical response to
pembrolizumab [9] (Fig. 4F), including differentially high expression of
genes CXCR6, STAT1, LAG3, CD8A, IDO1 and CCL5 in PD-L1 high tu-
mors.
R.J. Young, et al.
Fig. 3. Prognostic significance of intratumoral PD-L1+ immune cells according
to AJCC v8 ed. staging. Kaplan–Meier plots for OS for Stage I, II and III patients
for the PAH cohort. For (A-C) red line is high abundance of PD-L1+ ICs (≥5%);
black line is low abundance (< 5%).
5
Oral Oncology 101 (2020) 104516
Discussion
In the present study we have validated the prognostic significance of
PD-L1+ IT IC’s in HPV+OPSCC, utilising a large independent cohort to
confirm our previously published findings [6], and have identified
macrophages and cytotoxic T cells as the most commonly occurring PD-
L1+ ICs. We have also shown similar results to our previous findings
with abundance of CD8+ ICs, although with larger uncertainty due to
smaller number of events. As previously reported, no association was
found between tumoral PD-L1 expression and patient survival.
Characterisation of the PD-L1+ IT ICs, using mIHC, has shown the
majority of PD-L1+ IC’s to be macrophages (CD68+), cytotoxic T cells
(CD3+CD8+) or T helper cells (CD3+CD4+), with a smaller proportion
being B cells (CD19+) or dendritic cells (CD11c+). This is in keeping
with current knowledge of PD-L1 expression by immune cells [10,11].
In recent years there has been much interest surrounding immune
checkpoint-targeted immunotherapies which are showing impressive
and durable clinical responses in a variety of tumor types including
melanoma [12], non-small cell lung cancer [7] and HNSCC [13]. Our
gene expression data, showing over-expression of a number of immune-
related genes, including immune checkpoints such as LAG3, IDO1 and
TIM3 in tumors with high PD-L1 expression on IT ICs, may be more a
read out of an inflamed immune microenvironment, and may indicate a
group of patients who may respond favourably to immune therapies.
The key findings of this study are of importance for several reasons:
firstly, we have confirmed the identification of a subgroup of HPV
+OPSCC patients who have excellent long-term outcomes. Secondly,
addition of PD-L1+ IT IC score improves the prognostic ability of ex-
isting staging methods. Thirdly, two recent phase III trials in HPV
+OPSCC patients have shown poorer survival in patients treated with
radiotherapy and cetuximab compared with radiotherapy and cisplatin
[14,15], suggesting that there is still a need for improved identification
of low risk patients who would be suited to de-escalated treatment. Our
results suggest that patients with high abundance of PD-L1+ ICs may,
given their impressive survival, be an appropriate group for deintensi-
fication trials.
Our findings suggest that the abundance of CD8+ ICs may not be
robust enough to be used as a prognostic marker. We have recently
published work identifying CD103+CD8+ tissue resident memory cells
(TRMs) as being strongly prognostic for OS in the two cohorts described
in this study [16] and suggest that the identification of this important
subpopulation of CD8+ T cells warrants further investigation. We found
a significant association between the presence of PD-L1+ IT ICs and
CD103+ ICs (data not shown), both of which are likely surrogate bio-
markers of an active though restrained antitumor immune response.
This study aimed to validate our own previous work investigating
the prognostic role of PD-L1 and CD8 expressing ICs in HPV+OPSCC;
as such the technical elements of the study (antibodies and im-
munohistochemical methods used, scorers and scoring criteria), are all
consistent across the two studied cohorts. Limitations of the present
study include the use of only one PD-L1 antibody clone (sp142) when
several publications have demonstrated that different commonly used
PD-L1 clones give varying results when scoring tumor and immune cell
PD-L1 expression [17–19]. Rimm et al showed poor concordance be-
tween pathologists when scoring immune cell PD-L1 expression re-
gardless of the antibody clone used [19]. Whether our findings would
be replicated with different PD-L1 antibodies is unclear. We acknowl-
edge that at present, the use of PD-L1 immunohistochemistry as a
clinical or diagnostic biomarker still requires further refinement, and
there is still a need for reliable biomarkers that can be more readily
translated from the research setting to clinical diagnostic laboratories.
It may be that other IC biomarkers, such as CD103, with easier pa-
thological interpretation, may ultimately have better clinical applic-
ability than PD-L1.
Immune biomarkers show considerable promise for the stratifica-
tion of HPV+OPSCC patients and the selection of low risk candidates
R.J. Young, et al. Oral Oncology 101 (2020) 104516

Fig. 4. Multiplex immunohistochemistry and gene expression analysis. (A) Semi-quantitative image analysis of multiplex immunohistochemistry for PD-L1 and
immune cell markers CD3, CD8, CD4, CD68, CD19 and CD11c showed CD68+ macrophages and CD3+CD8+ cytotoxic T cells to be the most common PD-L1+
immune cells in intratumoral nests. (B-C) examples of colocalisation of PD-L1 expression (cyan) and immune cell markers; (B) CD68+ macrophages (purple) and (C)
CD8+ cytotoxic T cells (green), located within p16+ intratumoral nests (white). Arrows indicate PD-L1+ ICs. (D) Volcano plot showing differential expression of
immune related genes between HPV+OPSCC tumors with high abundance of PD-L1+ IT ICs compared to low abundance PD-L1+ IT IC tumors – upregulated genes in
PD-L1+ IT IC high tumors included PD1, IDO1, LAG3 and HAVCR2 (Tim3). (E) Heat map showing clustering of patient samples according to expression of a three
gene signature (INFG, CXCL9, CD274) previously described as being associated with response to atezolizumab. (F) Heat map showing clustering of patient samples
according to expression of a 16 gene ‘IFNγ related’ signature shown to be associated with response to pembrolizumab. For (E) and (F), red bars = high PD-L1, gray
bars = low PD-L1. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

for de-intensification trials. We validated the presence of PD-L1+ im-
mune cells within the intratumoral nests of HPV+OPSCC tumors as a
strongly favourable prognostic biomarker, although it’s broad applic-
ability as a diagnostic tool needs to be approached cautiously given the
potential issues associated with use of different PDL1 antibody clones
and reproducibility of scoring methods and interpretation of PD-L1
expression.
steering committees (all uncompensated), MSD, Merck, BMS, GSK,
Regeneron, Merck; BS – honoraria, BMS, AstraZeneca (Inst); consulting
or advisory role, BMS, MSD, AstraZeneca, Pfizer, Genentech, Eisai; in-
stitutional research funding, Pfizer, Roche; royalties from Veristrat
(Biodesix); UpToDate; travel, accommodation and expenses,
AstraZeneca, Roche, Merck, BMS, Novartis. All remaining authors have
declared no conflicts of interest.

Funding sources References

This work was supported by a grant from the Peter MacCallum
Cancer Centre Foundation and a NHMRC Career Development
Fellowship to BS.
Declaration of Competing Interest
The authors make the following declaration of competing interests,
all outside the submitted work: RY - institutional research funding,
Roche; SP - advisory boards, MSD, Merck; honoraria, Celgene; DU -
lecture fees, Bristol Myers Squibb (BMS); travel fees and lecture fees,
Merck Sharp & Dohme (MSD); travel fees and lecture fees, Roche; CC -
advisory boards, MSD (no honorarium); travel, accommodations and
expenses, Pfizer, MSD; PN – honoraria, BMS; institutional research
funding, BMS, Roche Genentech, Celgene-Juno, Advaxis, Compugen
and Allergan; DR - institutional research funding, MSD, Merck, BMS,
GSK, Roche, Regeneron, Amgen; Travel, MSD; advisory boards/trial
[1] Chaturvedi AK, D'Souza G, Gillison ML, Katki HA. Burden of HPV-positive or-
opharynx cancers among ever and never smokers in the U.S. population. Oral Oncol
2016;60:61–7.
[2] Chaturvedi AK, Anderson WF, Lortet-Tieulent J, Curado MP, Ferlay J, Franceschi S,
et al. Worldwide trends in incidence rates for oral cavity and oropharyngeal cancers.
J Clin Oncol 2013;31:4550–9.
[3] Rischin D, Young RJ, Fisher R, Fox SB, Le QT, Peters LJ, et al. Prognostic sig-
nificance of p16INK4A and human papillomavirus in patients with oropharyngeal
cancer treated on TROG 02.02 phase III trial. J Clin Oncol 2010;28:4142–8.
[4] Ang KK, Harris J, Wheeler R, Weber R, Rosenthal DI, Nguyen-Tan PF, et al. Human
papillomavirus and survival of patients with oropharyngeal cancer. N Engl J Med
2010;363:24–35.
[5] Huang SH, Xu W, Waldron J, Siu L, Shen X, Tong L, et al. Refining American Joint
Committee on Cancer/Union for International Cancer Control TNM stage and
prognostic groups for human papillomavirus-related oropharyngeal carcinomas. J
Clin Oncol 2015;33:836–45.
[6] Solomon B, Young RJ, Bressel M, Urban D, Hendry S, Thai A, et al. Prognostic
significance of PD-L1(+) and CD8(+) immune cells in HPV(+) oropharyngeal
squamous cell carcinoma. CancerImmunol Res 2018;6(3):295–304. Mar.
[7] Fehrenbacher L, Spira A, Ballinger M, Kowanetz M, Vansteenkiste J, Mazieres J,
et al. Atezolizumab versus docetaxel for patients with previously treated non-small-

6
R.J. Young, et al.
cell lung cancer (POPLAR): a multicentre, open-label, phase 2 randomised con-
trolled trial. Lancet 2016;387:1837–46.
[8] Kowanetz MZW, Mccleland M, et al. Pre-existing immunity measured by Teff gene
expression in tumor tissue is associated with Atezolizumab efficacy in NSCLC. J
Thoracic Oncol 2017;12:S1817–8.
[9] Ayers M, Lunceford J, Nebozhyn M, Murphy E, Loboda A, Kaufman DR, et al. IFN-
gamma-related mRNA profile predicts clinical response to PD-1 blockade. J Clin
Invest 2017;127:2930–40.
[10] Pollari M, Bruck O, Pellinen T, Vahamurto P, Karjalainen-Lindsberg ML, Mannisto
S, et al. PD-L1(+) tumor-associated macrophages and PD-1(+) tumor-infiltrating
lymphocytes predict survival in primary testicular lymphoma. Haematologica
2018;103:1908–14.
[11] Jacquelot N, Roberti MP, Enot DP, Rusakiewicz S, Ternes N, Jegou S, et al.
Predictors of responses to immune checkpoint blockade in advanced melanoma. Nat
Commun 2017;8:592.
[12] Larkin J, Chiarion-Sileni V, Gonzalez R, Grob JJ, Cowey CL, Lao CD, et al.
Combined Nivolumab and Ipilimumab or Monotherapy in Untreated Melanoma. N
Engl J Med 2015;373:23–34.
[13] Economopoulou P, Agelaki S, Perisanidis C, Giotakis EI, Psyrri A. The promise of
immunotherapy in head and neck squamous cell carcinoma. Ann Oncol
2016;27(9):1675–85. Sep.
Oral Oncology 101 (2020) 104516
[14] Gillison ML, Trotti AM, Harris J, Eisbruch A, Harari PM, Adelstein DJ, et al.
Radiotherapy plus cetuximab or cisplatin in human papillomavirus-positive or-
opharyngeal cancer (NRG Oncology RTOG 1016): a randomised, multicentre, non-
inferiority trial. Lancet 2019;393:40–50.
[15] Mehanna H, Robinson M, Hartley A, Kong A, Foran B, Fulton-Lieuw T, et al.
Radiotherapy plus cisplatin or cetuximab in low-risk human papillomavirus-posi-
tive oropharyngeal cancer (De-ESCALaTE HPV): an open-label randomised con-
trolled phase 3 trial. Lancet 2019;393:51–60.
[16] Solomon B, Young RJ, Bressel M, Cernelc J, Savas P, Liu H, et al. Identification of an
excellent prognosis subset of human papillomavirus associated oropharyngeal
cancer patients by quantification of intratumoral CD103+ immune cell abundance.
Ann Oncol 2019;30(10):1638–46. Oct 1.
[17] Hendry S, Byrne DJ, Wright GM, Young RJ, Sturrock S, Cooper WA, et al.
Comparison of Four PD-L1 Immunohistochemical Assays in Lung Cancer. J Thorac
Oncol 2018;13:367–76.
[18] Hendry S, Cooper WA, Fox SB. Response to editorial: the PD-L1 im-
munohistochemistry biomarker: two steps forward, one step back? J Thorac Oncol
2018;13:e185–6.
[19] Rimm DL, Han G, Taube JM, Yi ES, Bridge JA, Flieder DB, et al. A prospective,
multi-institutional, pathologist-based assessment of 4 immunohistochemistry assays
for PD-L1 expression in non-small cell lung cancer. JAMA Oncol 2017;3:1051–8.