0% found this document useful (0 votes)

165 views

Immune Landscape v11

Uploaded by

onco learn

We take content rights seriously. If you suspect this is your content, claim it here.

Available Formats

Download as PDF, TXT or read online on Scribd

0% found this document useful (0 votes)

165 views

Immune Landscape v11

Uploaded by

onco learn

We take content rights seriously. If you suspect this is your content, claim it here.

Available Formats

Download as PDF, TXT or read online on Scribd

You are on page 1/ 100

Surveying the

Immune Landscape of Cancer

Article Collection

37 Mass cytometry: a robust platform for

Identification of the immune checkpoint the comprehensive immunomonitoring
signature of multiple myeloma using of CAR-T-cell therapies
mass cytometry-based single-cell BY AURÉLIEN CORNEAU, CHRISTOPHE PARIZOT, MUSTAPHA CHERAI, EVE
analysis TODESCO, CATHERINE BLANC, ELENA LITVINOVA, STÉPHANIE NGUYEN,
DAMIEN ROOS-WEIL, AMÉLIE GUIHOT, FRANCOISE NOROL
BY JINHENG WANG, YONGJIANG ZHENG, CHENGGONG TU, HUI ZHANG,
British Journal of Haematology
KARIN VANDERKERKEN, ELINE MENU, AND JINBAO LIU
Clinical and Translational Immunology

53
COVER IMAGE © STANDARD BIOTOOLS

A phase1b/2 study of azacitidine with

PD-L1 antibody avelumab in relapsed/
refractory acute myeloid leukemia
BY KAPIL SAXENA, SHELLEY M. HERBRICH, NAVEEN PEMMARAJU, TAPAN
M. KADIA, COURTNEY D. DINARDO, GAUTAM BORTHAKUR, SHERRY A.
PIERCE, ELIAS JABBOUR, SA A. WANG, CARLOS BUESO-RAMOS, SANAM
LOGHAVI, GUILLIN TANG, CORA M. CHEUNG, LYNETTE ALEXANDER,
STEVEN KORNBLAU, MICHAEL ANDREEFF, GUILLERMO GARCIA-MANERO,
FARHAD RAVANDI, MARINA Y. KONOPLEVA, AND NAVAL DAVER
Cancer

3
Introduction
The immune system has emerged as a key player in the First, Pisano et al. (2021) utilized a preclinical model
direction of cancer progression, displaying both pro- and of ovarian cancer to examine the immune landscape
antitumor capabilities as well as the capacity to be leveraged of the ascitic fluid surrounding the tumor and tumor
for treatment. To fully tease apart the interaction of immune nodules. Mass cytometry was able to identify both pro-
cells within tumor microenvironments, researchers need to and antitumor activity of immune cells in the ascitic
be able to clearly identify the presence and state of immune fluid and cell populations in the tumor nodules that
subpopulations. Conventional flow cytometry has been could serve as targets for therapeutic intervention.
a mainstay of immunophenotyping for decades. Spectral
Eiva et al. (2022) further expanded on the use of CyTOF
flow cytometry, a relatively new approach to fluorescence for examining ovarian cancer by analyzing biomarkers for
cytometry, is also being used for immunophenotyping tumor infiltrating lymphocytes (TILs) and their origin in
studies. However, both approaches suffer in the number of tumor tissue and ex vivo cells. Three identifiable populations
cell markers that can be incorporated into a single panel of TILs were found to differentiate from a subset of CD137
due to issues of emission spectra overlap. Mass cytometry expressing TILs, thereby indicating that CD137 may be of use
or cytometry by time-of-flight mass spectroscopy (used as a selective biomarker identifying tumor-specific TILs.
in CyTOF® instruments) has emerged as a critical tool in
the identification and monitoring of immune cells in the Wang et al. (2020) used CyTOF to gain a better
research and clinical spaces due to the precision of signal understanding of the immune checkpoint signals
detection and lack of anything analogous to spectral overlap. associated with multiple myeloma (MM). They discovered
a distinct signature pattern of increased activated CD4
In contrast to fluorescence cytometry, cells for CyTOF and CD8 T cells, and CD8+ natural killer T-like and NK
analysis are stained with antibodies labeled with heavy cells in bone marrow of MM patients, providing new
metal isotopes not endogenous to biological systems. targets for immune checkpoint blockade therapy.
Labeled cells are injected as single-cell droplets and passed
Last, Saxena et al. (2021) presented a clinical trial on the
through a hot inductively coupled plasma to ionize the
safety and efficacy of combining an anti-PD-L1 checkpoint
isotope labels of bound antibodies. The ionized cloud
inhibitor (avelumab) with a hypomethylating agent
enters the time-of-flight chamber and, based on the label’s
(azacytidine) in the treatment of relapsed/refractory
atomic mass, the time to reach the detector identifies the
acute myeloid leukemia (AML). This method of treatment,
metal isotopes and associated antibody label with high
which primarily targets the PD-L1 ligand, has displayed
precision. By virtue of this labeling and identification
limited ability to treat AML. Mass cytometric analysis of
technology, CyTOF can incorporate 50 or more cellular
bone marrow samples from study participants treated
markers in a single panel with negligible overlap in with this combination uncovered a previously unknown
detection. Therefore, CyTOF is ideal for immunological activity of another PD-1 receptor ligand (PD-L2) that
studies and monitoring of therapeutic efficacy in clinical could potentiate immune evasion or tolerance. This
research. Key benefits are sensitive immunophenotyping suggests that blockade of PD-1 receptor itself instead
assays where fluorescence spillover and autofluorescence of the ligand may be a more effective treatment.
are eliminated as sources of background signal and
less precious sample is required for analysis. While all of the above articles provide examples of the
use of mass cytometry for immune monitoring and
In this article collection, we provide examples of how CyTOF characterization, recent advances in the technology stand
technology allows for a highly precise and informative to give researchers and clinicians an even deeper and more
interrogation of immune cells in response to clinically reproducible ability to understand immune system activity.
relevant treatment approaches. To begin, we present four The CyTOF XT™ system is the next generation of mass
articles illustrating examples of using mass cytometry to cytometry technology, automating many aspects of the
monitor immune responses to cancer and effectiveness CyTOF workflow to increase throughput and reproducibility.
of therapeutic interventions in research studies. Further, when combined with the Maxpar® Direct™ Immune

4
Profiling Assay™, the use of high-parameter cytometry for
immune studies that are highly reproducible is facilitated.
References
The full advantages of the system and supporting data Pisano, S. et al. (2021). Assessment of the immune landscapes
can be found in the provided application note titled of advanced ovarian cancer in an optimized in vivo model. Clin
CyTOF XT: The Next Generation of Mass Cytometry. Transl Med; 11: e551. https://doi.org/10.1002/ctm2.551
Importantly, mass cytometry has demonstrated multi-site Eiva, M.A. et al. (2022). Systematic analysis of CD39, CD103,
reproducibility. To support this, we provide two additional CD137, and PD-1 as biomarkers for naturally occurring tumor
publications utilizing CyTOF with the Maxpar Direct antigen-specific TILs. Eur J Immunol; 52: 96–108. https://doi.
Immune Profiling Assay. Bagwell et al. (2020) reported the org/10.1002/eji.202149329
findings of a multi-site study for validating the use of the
assay in identifying immune cell populations from whole Wang, J. et al. (2020). Identification of the immune checkpoint
blood and peripheral blood mononuclear cell samples. signature of multiple myeloma using mass cytometry-based
Each site utilized the same instrumentation and assay single-cell analysis. Clin Transl Immunol; 9: e1132. https://doi.
on the provided samples. The study concluded that the org/10.1002/cti2.1132
Maxpar Direct Immune Profiling Assay can robustly and
reproducibly detect 37 distinct immune cell populations Saxena, K. et al. (2021). A phase 1b/2 study of azacitidine with
using a 30-marker dry-format panel. Finally, Corneau et PD-L1 antibody avelumab in relapsed/refractory acute myeloid
al. (2021) detailed the customization and validation of a leukemia. Cancer; 127: 3761–71. https://doi.org/10.1002/
Maxpar Direct Immune Profiling Assay based panel for CAR cncr.33690
T cells and non-CAR cells in peripheral blood to monitor
the outcomes of CAR T cell therapies in development. CyTOF XT: The Next Generation of Mass Cytometry. Standard
BioTools Application Note
With the ever-increasing complexity of the immune system
becoming apparent during the development of therapeutic Bagwell, C.B. et al. (2020). Multi-site reproducibility of a human
interventions, methods to clearly identify and track the immunophenotyping assay in whole blood and peripheral
activity of immune cell types are crucial to determining blood mononuclear cells preparations using CyTOF technology
their effectiveness. Mass cytometry stands to revolutionize coupled with Maxpar Pathsetter, an automated data analysis
the ability of researchers and clinicians to monitor immune system. Cytometry; 98: 146–60. https://doi.org/10.1002/
responses to therapies in a precise manner not afforded cyto.b.21858
by conventional flow cytometry. By providing examples of
applied mass cytometry for immune monitoring, we hope Corneau, A. et al. (2021). Mass cytometry: a robust platform for
to educate readers on this powerful tool and how it may be the comprehensive immunomonitoring of CAR-T-cell therapies.
leveraged for their specific translational and clinical research Br J Haematol; 194: 788–92. https://doi.org/10.1111/bjh.17551
goals. For more information and resources for CyTOF, we
encourage you to visit the Standard BioTools Maxpar Direct
Immune Profiling System and CyTOF XT web pages. For Research Use Only.
Not for use in diagnostic procedures.
By Jeremy Petravicz, PhD, Editor,
Current Protocols

5
Received: 14 January 2021 Revised: 6 August 2021 Accepted: 9 August 2021

DOI: 10.1002/ctm2.551

RESEARCH ARTICLE

Assessment of the immune landscapes of advanced ovarian

cancer in an optimized in vivo model

Simone Pisano1,2 Stefania Lenna1 Gareth D. Healey2 Fereshteh Izardi2

Lucille Meeks1 Yajaira S. Jimenez1,3 Oscar S Velazquez1
Deyarina Gonzalez2 Robert Steven Conlan1,2 Bruna Corradetti1,2,3

1Department of Nanomedicine, Houston

Methodist Research Institute, Houston, Abstract
Texas Background: Ovarian cancer (OC) is typically diagnosed late, associated with
2Center for NanoHealth, Swansea high rates of metastasis and the onset of ascites during late stage disease. Under-
University Medical School, Swansea, UK
3
standing the tumor microenvironment and how it impacts the efficacy of current
Texas A&M Health Science Center,
College of Medicine, Bryan, Texas treatments, including immunotherapies, needs effective in vivo models that are
fully characterized. In particular, understanding the role of immune cells within
Correspondence
the tumor and ascitic fluid could provide important insights into why OC fails to
Bruna Corradetti, Department of
Nanomedicine, Houston Methodist respond to immunotherapies. In this work, we comprehensively described the
Research Institute, 6670 Bertner Ave., immune cell infiltrates in tumor nodules and the ascitic fluid within an opti-
Houston, TX 77030.
Email: [email protected]
mized preclinical model of advanced ovarian cancer.
Methods: Green Fluorescent Protein (GFP)-ID8 OC cells were injected
Funding information intraperitoneally into C57BL/6 mice and the development of advanced stage OC
European Union’s Horizon 2020 Research
and Innovation Program, Grant/Award monitored. Nine weeks after tumor injection, mice were sacrificed and tumor
Number: 663830 nodules analyzed to identify specific immune infiltrates by immunohistochem-
istry. Ascites, developed in tumor bearing mice over a 10-week period, was char-
acterized by mass cytometry (CyTOF) to qualitatively and quantitatively assess
the distribution of the immune cell subsets, and their relationship to ascites from
ovarian cancer patients.
Results: Tumor nodules in the peritoneal cavity proved to be enriched in T cells,
antigen presenting cells and macrophages, demonstrating an active immune
environment and cell-mediated immunity. Assessment of the immune landscape
in the ascites showed the predominance of CD8+ , CD4+ , B– , and memory T
cells, among others, and the coexistance of different immune cell types within
the same tumor microenvironment.
Conclusions: We performed, for the first time, a multiparametric analysis of the
ascitic fluid and specifically identify immune cell populations in the peritoneal
cavity of mice with advanced OC. Data obtained highlights the impact of CytOF

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the
original work is properly cited.
© 2021 The Authors. Clinical and Translational Medicine published by John Wiley & Sons Australia, Ltd on behalf of Shanghai Institute of Clinical Bioinformatics

Clin. Transl. Med. 2021;11:e551. wileyonlinelibrary.com/journal/ctm2

6
https://doi.org/10.1002/ctm2.551
PISANO et al.

as a diagnostic tool for this malignancy, with the opportunity to concomitantly

identify novel targets, and define personalized therapeutic options.

KEYWORDS
ascites, CyTOF, immunotherapy, mass cytometry, model, ovarian cancer, peritoneal cancers

1 INTRODUCTION total absence) of CD8+ T cells within the TME,12 or the

inability of dendritic cells (DCs) to effectively present
Ovarian cancer (OC) is the 7th most common cause of antigen and stimulate a cytotoxic response.13 One of the
death in women worldwide, with over 21 000 new cases possible mechanisms behind DC inactivation has been
expected in the United States in 2020.1 Survival rates vary provided by Cubillos-Ruiz et al.14 The authors demon-
according to the stage of disease, with a 5-year survival rate strated that the reduced capability of DC to support an
of around 30% for advanced cancers,2,3 the most common anticancer immune response is associated to the tran-
of which is high-grade serous ovarian carcinoma (HGSOC) sient, yet abnormal lipid accumulation in the endoplas-
accounting for more than 50% of cases. Unfortunately, only mic reticulum, which obstructs their normal antigen-
33% of OC cases are identified early, the majority being presenting capacity.14 Another factor proposed to play a
diagnosed at a later, more advanced stage, and associated role in ovarian cancer progression at advanced stages and
with a significantly worse prognosis.4 resistance to immunotherapy is the presence of transform-
According to the National Comprehensive Cancer Net- ing growth factor-β (TGF-β). Specifically, TGF-β is a potent
work (NCCN) guidelines, the standard of care ther- immunosuppressor within the tumor environment being
apy for HGSOC involves debulking surgery followed involved in several tumor-associated processes, includ-
by platinum- or taxol-based chemotherapies. Among ing the increase of the epithelial to mesenchymal transi-
other recommended treatments, liposomal doxorubicin tion, the promotion of angiogenesis and immune suppres-
is a viable option for both early and advanced-stage sion. The enhanced secretion of TGF-β within the tumor
disease.5 Targeted therapeutic approaches, recently added microenvironment is associated to the recruitment of reg-
to standard clinical practice, provide improved sur- ulatory T cells via expression of FoxP3, which ultimately
vival rates and include: vascular endothelial growth fac- results in diminished cytotoxic T-lymphocytes and in a
tor (VEGF)-A inhibitors,6 and poly (ADP-ribose) poly- reduced presence of DCs.15,16 There is, however, a paucity
merase (PARP) inhibitors, which are indicated for patients of evidence on the specific roles of immune cell popula-
with a BReast CAncer gene (BRCA1/2) mutation.5 OC tions within the OC TME. Hence, a more comprehensive
remains a complex disease to treat, owing to the high understanding of the immune cell landscape would pro-
chemotherapy-resistance emergence rate,7 and in recent vide an important platform for the development of more
years great emphasis has been placed on the employ- efficacious immunotherapeutic strategies.
ment of immunotherapies to combat this issue, although The accumulation of fluid within the peritoneal cav-
currently no clinically approved immunotherapy for ity (ascites), which contains a variety of soluble and
HGSOC exists. Modest activity within recurrent OC cellular components, is characteristic of advanced stage
patients (which included epithelial, fallopian, or pri- OC. Indeed, more than one third of OC patients present
mary peritoneal OC) has been reported in the Phase II with ascites at diagnosis, which has been correlated with
KEYNOTE-100 study for the checkpoint inhibitor (CPI) its spread within the peritoneal cavity and poor patient
Pembrolizumab.8 Additionaly, several Phase III trials are prognosis.17 The accumulation of ascites occurs as a con-
exploring the combination of CPI with PARP or VEGF sequence of unbalanced drainage of the peritoneal cav-
inhibitors to determine any therapeutic synergies.9 ity, due to obstruction of the lymphatic system by cancer
Limited immunotherapy efficacy observed to date, how- cells,17 or by increased leakage of fluid from the microves-
ever, could be explained by the typically “cold” immune sels lining the peritoneum.18 Ascites build-up also con-
status of OC. Indeed, the advanced OC tumor microen- tributes to malignant progression by facilitating multifocal
vironment (TME) is characterized by a lack tumor infil- cancer cell dissemination on the peritoneal surface.19 The
trating lymphocytes (TILs) and failed T-cell priming due presence of an intraperitoneal ascitic current, which acts
to a combination of poor antigen presentation and an as a means of transport of OC spheroids, further facilitates
intrinsic insensitivity to T-cell killing.10,11 More specifi- peritoneal, lymphatic, and hematogenous metastasis,20 a
cally, tumor growth is associated with a scarcity (if not phenomenon that falls within the multistep process of

7
PISANO et al.

metastatic dissemination. Soluble and cellular compo- 2.2 Lentivirus transduction, lentiviral
nents within the ascitic fluid have also been shown to influ- infection of ID8 cells with the luciferase
ence metastatic behavior.17 Soluble components, includ- vector and cell line selection
ing growth factors, cytokines, chemokines, and extracel-
lular matrix pieces, inhibit T helper cell proliferation21 The ID8-Luc/GFP cell line was generated by transduction
and DC maturation22 mediated by IL-10. Cellular compo- with Lentivirus particles containing the CMV promoter
nents, such as resident tumor cells or tumor-associated for the expression of humanized firefly luciferase (hLUC)
fibroblasts, or nonresident immune cells, on the other and the SV40 promoter for the expression of GFP pro-

Briefly, ID8 cells were plated at 2 × 104 cells per well (12-
hand, have a wide ranging impact on the TME. The tein according to manufacturer’s protocol (GeneCopoeia).
presence, functionality, and effect of specific, singularly
taken immune cell populations within the ascitic fluid has well plate, Corning) and incubated overnight at 37◦ C in a
been widely described, unraveling the association between humidified 5% CO2 atmosphere. Cells were then infected
the presence of tumor-infiltrating CD8+ T cells and the with 10 MOI of Lenti-PAC™ plasmid mix (GeneCopoeia
prolonged disease-free survival,23 or unmasking the role Inc.) in the presence of 8 μg/mL polybrene (Sigma). After
of T regulatory cells in creating an immunosuppressive overnight incubation at 37◦ C/5% CO2 , the viral super-
environment.24 As such, ascites represents a potentially natant was discarded, and cells were washed with 1×
very informative source of information regarding the effect PBS (ThermoFisher) prior to the addition of warmed HG-
of immune cells on metastatic disease progression. More- DMEM media. Three days after infection, cells with high
over, its presence in over 30% of patients at diagnosis ren- levels of GFP expression were selected by Cell Sorter NIR
ders it an important issue to tackle and explore. Hence, a Aria II (BD Bioscience) and expanded for a week in HG-
complete profiling of the ascites immune content would DMEM media in presence of 1 μg/mL puromycin (Invitro-
prove useful if done on patients in a tailored fashion. How- gen) to further select transfected cells and generate a stable
ever, fundamental research on the biological interactions cell line.
of the components of advanced OC ascites requires reliable
in vivo models.
In this study, we optimize the development of an 2.3 In vivo propagation of ID8-GFP
advanced OC model in immunocompetent mice to fill the tumors
gap in the understanding of the immune landscape within
the peritoneal cavity. For the first time, we apply mass Female C57BL/6 (5-6 weeks old) were purchased from
cytometry to comprehensively describe the immunological the Charles Rivers laboratories. All animal studies were
TME within the ascites and to provide insights about the carried out in accordance with guidelines determined by
effectiveness of the selected preclinical model in reproduc- the Animal Welfare Act and the Guide for the Care and
ing the human tumor immunomicroenvironment. Finally, Use of Laboratory Animals and complied with protocols
we propose mass cytometry as an accurate strategy for the approved by the Institutional Animal Care and Use Com-
development of personalized strategies against advanced mittee at the Houston Methodist Research Institute (AUP-

3 groups (n = 5 mice per group) and injected intraperi-

OC and all cancers metastasizing within the peritoneal 0219-0013). Briefly, C57BL/6 female mice were divided into

toneally with 5 × 106 , 1 × 107 , or 1.5 × 107 ID8-Luc/GFP cells

cavity.

in 200 μL of PBS. Cells were injected into the lower right

2 METHODS quadrant of the abdomen. Mice weights (g) were recorded
daily following ID8-Luc/GFP cell injection and plotted as
2.1 Cell line fold change. Representative macroscopic images of tumors
and ascites development were taken with a smartphone
The ID8 cell line, originated from mouse ovarian sur- camera.
face epithelial cells (MOSEC), was purchased from Merck-
Millipore. Cells were cultured in High Glucose Dulbecco’s
Modified Eagle medium (HG-DMEM) (Sigma) supple- 2.4 Bioluminescence, imaging, and
mented with 10% fetal bovine serum (FBS, ThermoFisher), tumor localization within the abdominal
5 μg/mL insulin, 5 μg/mL transferrin and 5 ng/mL sodium cavity
selenite (1× ITS, Sigma) and 1× Penicillin-Streptomycin
Solution (Sigma). Culture conditions were 37◦ C in a To track tumor growth, luciferase luminescence was
humidified 5% CO2 atmosphere. detected using a Xenogen IVIS Spectrum imaging system

8
PISANO et al.

(PerkinElmer) as previously described.25 Briefly, 200 μL T A B L E 1 Panel of the antibodies selected for mass cytometry,
of 15 mg/mL D-luciferin was injected into the mice CyTOF. List of the 33 antibodies used, and their metal conjugation
abdomen and the bioluminescent signal evaluated after Target Metal tag
10 min to obtain the peak photon emission per sec- CD45 Pr141
ond. The signal was quantified using the Living Image MHC II Nd142
software (PerkinElmer) and the total photon flux emis- CD11b Nd143
sion (photons/second) in the regions of interest (ROI) Ly6C Nd144
recorded, starting at day 8 after tumor cell injection.
Ly6G Nd145
Images were normalized using the Living Image software
F4/80 Nd146

1.7 × 104 and 9.7 × 104 photons/s, respectively.

(PerkinElmer) with a minimum and maximum radiance of
CD11c Sm147
CD38 Nd148
Arg-1 Sm149
2.5 Hematoxylin and eosin (H&E) SiglecF Nd150
staining and immunohistochemistry (IHC) CD206 Eu151
CD62L Sm152
Sixty-three days after ID8-Luc/GFP cell injection, mice CD103 Eu153
had a strong tumor signal intensity by IVIS. Hence, mice iNOS Sm154
were sacrificed and the peritoneal membrane, abdominal PD-L1 Gd155
tumors, and liver were sampled, fixed in 4% paraformalde- TNFa Gd156
hyde solution overnight, and embedded in paraffin. Paraf- CD64 Gd158
fin embedded tissues were subsequently sectioned at a
TCRgd Tb159
thickness of 5 μm and hematoxylin and eosin (H&E) stain-
Foxp3 Gd160
ing performed to enable general inspection of the tissues.
RORgt Dy161
The 5 μm thick sections were also used for immunohisto-
CD8α Dy162
chemical staining. Sections were incubated with primary
anti-CD3 (rabbit, Dako), anti-MHC-II (rat, eBioscience), or Tbet Dy163
anti-F4/80 (rat, BioRad) for 1 h at room temperature (RT) CD25 Dy164
in a moist chamber. Sections were imaged with a EVOS R IFN-γ Ho165
FL Auto Imaging System (Life Technologies). CD44 Er166
CD86 Er167
CD80 Er168
2.6 Ascites extraction and mass PD-1 Tm169
cytometry by time of flight (CyTOF) B220 Er170
analysis NK1.1 Yb171

Seventy days after tumor cell injection with 1 × 107

CD19 Yb173
CD4 Yb174
ID8-Luc/GFP cells, mice started developing ascitic fluid.
TCR β Lu175
Ascites onset was detected by abdomen palpation, by
eye and by weight increase. After reaching a weight of
30 g, three mice were sacrificed, and the ascetic fluid
collected by syringe suction following abdominal incision. by incubation with 25 μM cisplatin for 5 min. At these
Ascitic fluid was centrifuged, and red blood cells lysed conditions, cisplatin preferentially reacts with proteins in
by incubation in Ammonium-Chloride-Potassium (ACK) dead cells and it widely established as a viability reagent
lysing buffer (ThermoFisher) for 10 min at RT. Immune for mass cytometry.26 After washing in Maxpar R Cell
cell enrichment was achieved using Percoll gradient Staining Buffer (Fluidigm), cells were resuspended in 40
centrifugation. Briefly, the cell pellet was resuspended μL surface-staining antibody (Ab) mix and incubated at

onto 50% Percoll and centrifuged at 620 × g without the

in 85% Percoll (GE Healthcare), then carefully layered RT for 30 min. Antibodies were purchased from Biolegend
(except for Arginase-1 and NOS2, which were purchased
brake for 30 min at 4◦ C. After centrifugation, three layers from eBioscience) and conjugated to the metals using
of cells were present. The middle layer, consisting of the the Maxpar R X8 Multimetal Labeling Kit (Fluidigm).
immune cells of interest, was recovered and used for Selected Ab are shown in Table 1. Cells were then washed
mass cytometry staining. Cell viability was determined 2× and fixed with 100 μL of Fix/Perm buffer (eBioScience)

9
PISANO et al.

for 10 min, followed by the addition of 200 μL Perm buffer week 1 (P < 0.001). No differences in tumor growth were
(eBioScience) for 10 min. The intracellular staining was noted between the three treatment groups, although all
performed by diluting cells in 50 μL Ab mix and incubat- three groups showed increased tumor size over time. Rep-
ing them at RT for 60 min. After the washing steps, the resentative pictures of the signal produced by the tumor
cell ID DNA intercalator (500 μM, Fluidigm) was added within the abdomen are shown in Figure 1B. Figure 1C
to cells in a 1:1000 dilution for 30 min at RT. Cells were shows the fold variation of mice weights over the 9-week
then washed, counted, and filtered through blue-capped period, indicating a correlation between tumor growth and

injection of 1 × 107 or 1.5 × 107 cells. Statistical difference

tubes (35 μm) before resuspension in 50 μl deionized mouse weight, which was particularly evident following
water and the addition of 50 μl of EQ-beads (eBioScience).

pared to the day of injection (P < 0.001). This difference

Samples were acquired by Helios CyTOF machine. A total was seen within groups at different time points when com-

became more marked after day 23 days in 5 × 106 group and

of 100 000 events were recorded for each sample, and

after 37 days in 1 × 107 group. No statistically significant

subsequently analyzed on Cytobank. Immune cell pop-

intragroup differences were recorded in 1.5 × 107 group.

ulations were identified by manual gating. The intensity
of the signal in the viSNE plots obtained was divided into
three main groups. The following thresholds were used for Sixty-three days after tumor cell injection, mice were sac-
categorization of the immune cell subtypes and applied rificed and organs extracted to better localize the tumors
to each specific viSNE plot: if the majority of the region within the abdomen. ID8-Luc/GFP cells were identified
was in the high range of expression (the red colored area, in several abdominal organs including liver, kidneys, and
with numerical values varying according to the analyzed spleen, and multiple tumor nodules/formations were ran-
marker), the marker was considered highly expressed domly distributed around the abdominal cavity within the
(++); if the region was in the middle range of expression peritoneum (Figure 1D).
(color-coded azure to orange) then the marker was con-
sidered to be moderately lowly expressed (+). Finally, if
the area exhibited mostly lower expression (indicated by 3.2 Histological analysis of liver and
blue and dark blue colors), markers were considered not tumor nodules
expressed (−).
Tumors, extracted at day 63 after intraperitoneal injec-
tion, were processed and stained with H&E. Tumor growth
2.7 Statistical analysis occurred in two main areas; firstly the inner surface of the

ible in mice from the 1 × 107 group (Figure 2A), and second,
peritoneal membrane, as can be inferred from nodules vis-
Statistical analysis was performed by ANOVA for all
experiments that required it. More specifically, a two-way the abdomen, where multiple tumor masses were found in
ANOVA with post hoc Dunnett comparisons to week 1 or multiple organs of the lower abdomen (Figure 2B). H&E

analyses, respectively. Data with a P < 0.05 were consid-

Day 0 was employed for tumor signal and tumor weight staining of the liver, peritoneal membrane, and tumor

ered significant (∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001). All
masses within the abdominal cavity shows the presence of
tumor growth within all three treatment groups and illus-

expressed as the mean ± standard deviation (SD).

results were obtained from independent experiments and trates the extent of tumor infiltration throughout the peri-
toneal cavity (Figure 2C).

3 RESULTS 3.3 Immunohistochemistry of immune

infiltrates within tumors
3.1 Tumor model optimization

After ID8-Luc/GFP cells (5 × 106 , 1 × 107 or 1.5 × 107 ) were

Tissue sections were also analyzed by immunohistochem-
istry (IHC) to identify immune cell infiltration within
intraperitoneally injected, tumor growth was assessed over tumors found on the peritoneal membrane. The presence

1 × 107 cells showed a 13.6 ± 9 fold increase in tumor

a 9-week period. After 9 weeks, animals injected with of cell surface markers for T cells, antigen presenting
cells (APC), and macrophages (CD3, MHC-II, and F4/80,

1 (P < 0.001), as quantified by IVIS (Figure 1A). Mice

growth, based on fluorescence signal, compared to week respectively) was investigated in mice from all treatment

injected with 5 × 106 or 1.5 × 107 cells showed a 8.2 ± 4

groups (Figure 3). Significant immune cell infiltration was

fold and 5.6 ± 43.9 fold increase in tumor growth, respec-

apparent in tumors present in all the treatment groups. Of
note, there were high numbers of T cells and macrophages,
tively, again based on fluorescence signal, compared to indicating an active immune environment.

10
PISANO et al.

injection of 5 × 106 , 1 × 107 , or 1.5 × 107 cells (n = 5). (B) Representative IVIS images of each tumor group taken after 6 weeks from tumor cells
F I G U R E 1 Tumor model generation and optimization. (A) Tumor growth signal quantification over a 9-week period by IVIS following

injection. (C) Mice weights (grams), expressed as fold change, during the 9-week experimental period (n = 5). (D) IVIS images of organs
extracted from the mice abdomen 9 weeks after injection with 5 × 106 , 1 × 10 7 , or 1.5 × 107 ID8-Luc/GFP cells. Extracted organs include liver,
spleen, kidneys, lungs, heart, peritoneal membranes, tumor nodules. Epi-fluorescent signals reflects the presence of ID8-Luc/GFP cells. Data

test; values differ from week 1 (A) or day 0 (C), *P < 0.05, **P < 0.01, ***P < 0.001
are expressed as mean (SD) from 5 independent experiments. Data were analyzed by ANOVA and Dunnett’s pairwise multiple comparison

11
PISANO et al.

F I G U R E 2 Histological assessment and localization of tumor nodules within the peritoneal cavity. (A) Representative images of tumor
growth on the peritoneal membrane (green arrows). (B) Representative images of tumor growth within the lower abdominal cavity, indicated
by the green arrow in the inset image. (C) H&E staining of tumor nodules found in the liver, peritoneal membrane, and abdominal cavity
(black arrows). Magnification: 10×, scale bar: 400 μm

A similar trend of immune cells recruitment to the within the ascites of tumor-bearing mice were analyzed
tumor masses was observed in the cancerous fragments by mass cytometry. Seventy days after tumor cell injection,
extracted from different areas of the abdominal cavity. Fig- ascitic fluid formation resulted in a swollen abdomen
ure 4 shows a strong presence of immune infiltrates despite that was apparent and palpable (Figure 5A). Differen-
the varied histological landscapes of the tissues examined. tial expression analysis of specific immune cell surface
markers (Table 1) present on CD45+ cells within ascitic
fluid was performed through mass cytometry (CyTOF)
3.4 Immune characterization of ascites analysis. Results from this analysis are plotted onto a
through mass cytometry (CyTOF) viSNE graph (Figure 5B) that plots CD45+ cells on a
two-dimensional map and identifies individual cells by
To obtain a better understanding of the immune landscape their expression of the specific immune cell markers cho-
of our metastatic OC model, immune cell populations sen (for gating strategy, see Figure S1). From these data,

12
PISANO et al.

F I G U R E 3 Identification of immune cell infiltration within tumors of the peritoneal membrane. Representative IHC images of each of
the experimental groups showing the presence of CD3+, MHC-II+, and F4/80+ cells (T cells, APC, and macrophages, respectively) within
tumors on the peritoneal membrane. Black arrows indicate the tumor masses. Magnification: 20×, scale bar: 200 μm

and CTL cells (4.6% ± 3.1% and 4.85% ± 3.4%, respectively)

immune cell population percentages and numbers can be of CD8+ T cells, while the relative percentages of T helper
derived (Figure 5C) enabling a comprehensive analysis
of the ascitic fluid immune cell population. Within the were similar.
present study, the most abundant cell populations in the Table 2 shows the list of all markers identified in each

± 9.6%), CD8+ T cells (38.5% ± 4.5%), and CD4+ T cells

ascitic fluid of tumor-bearing mice were B cells (27.3.6% immune cell population and differentiates them based

(20.7% ± 3.5%), with myeloid immune cells, including

on the level of expression (see Section 2.6). For instance,
the B-cell group is characterized by the high expression
monocytes, macrophages, DCs, eosinophils, and neu- of the markers CD38, B220, MHC-II, CD19, and CD80.
trophils accounting for the remaining 15% of the total cell In addition, CD25, PD-L1, and PD-1 are found to be
population. expressed by both B cells and CD4+ and CD8+ cells at
Further characterization revealed the presence of spe- high and low expression levels, respectively. CD4+ and
cific subpopulations among CD8+ and CD4+ T cells (Fig- CD8+ cells share high expression levels of TCRβ and
ure 5D). Specifically, based on their expression of specific IFN-γ (in addition to the cell specific markers CD4 and

T cells (Ly6C+/CD44+, 23.1% ± 12.1%), (ii) T helper cells

markers we identified the presence of (i) memory CD8+ CD8α, respectively), with Ly6C, Tbet, CD25, and CD103

(IFNγ+/CD4+, 22.2% ± 15.3%), and (iii) cytotoxic T lym-

being specifically expressed by the CD8+ population. Treg

phocytes (CTL, IFNγ+/CD8+, 12.5% ± 9.05%). In addition,

cells express high levels of Foxp3, CD25, CD4, and CD44

low expression of PD1 in 8.03% ± 7.4% of CD4+ T cells and

markers among others, with CD44 being also present in

2.3% ± 1.5% of CD8+ T cells was noted indicating a low

eosinophils, NK cells, monocytes, macrophages, DCs, and
neutrophils. Macrophages show high expression of F480
level of T-cell exhaustion. The percentages of remaining and share with monocytes the moderate to low expression

were 61.9% ± 24.3% and 69.7% ± 28.5%, respectively. The

CD8+ and CD4+ T cells with no identified subpopulations of CD11b, TCRgt, CD64, and CD80. Similarly, the presence
of CD11b, Ly6C, and Ly6G is observed in both, eosinophils,
CD8+/CD4+ T-cell ratio was 1.65, indicating a prevalence and neutrophils.

13
PISANO et al.

F I G U R E 4 Identification of immune cell infiltrates within abdominal cavity tumors. Representative IHC images of CD3+, MHC-II+,
F4/80+ cells (T cells, APC, and macrophages, respectively) within tumor nodules (indicated by black arrows) found within the abdominal
cavity. Magnification: 20×, scale bar: 200 μm

To complement the information provided in Table 2 cell line) is widely used to generate preclinical models of
and visualize the immune cell marker expression, data advanced OC. This is due to its capacity to closely repro-
from the CyTOF experiments were arranged into sub- duce the histopathological nuances that are characteris-
groups according to marker expression and their associ- tic of patients with advanced OC. These include tumor
ation with specific cell types (Figure 6). Several immune dissemination across the peritoneal cavity, a specific pat-
markers were associated with more than one cell popula- tern of invasion and the formation of ascites, which fur-
tion; however, the three main cell types identified were: ther increases the metastatic process and hinders therapy
B cells; CD4+ and CD8+ T cells; neutrophils, eosinophils, effectiveness.29 In addition, ID8 cells have been found to
macrophages, monocytes, and dendritic cells. The expres- express Pax8,30 a member of a transcription factor family
sion of CD62L, iNOS, and CD206 was not apparent on any that has also been linked to a role in OC development.31
cell population. Lastly, the immunological nature of ID8-based OC models
further increases their clinical relevance as they also allow
for the testing of innovative immunotherapies with the
4 DISCUSSION potential to treat this malignancy. As such, ID8 cells have
been widely used to test different hypotheses. For example,
The treatment of advanced OC is challenging, espe- Wilson et al used the ID8 model to track nuclear factor-
cially considering the altered physical transport properties kappa B (NF-κB) signaling during cancer progression,32
that create an immunosuppressive environment27,28 and while Zhang et al used ID8 cells stably expressing the vas-
limit responsiveness to current immunotherapy strategies. cular endothelial growth factor to demonstrate increased
Efforts to optimize animal models that comprehensively tumor-progression rate and ascites formation.33 However,
mimic cancer development in vivo and thus enable new some doubts about the stability of this cell line have been
therapeutic strategies to be tested are ongoing. In this con- raised, since the onset of ascites has been reported to alter
text, the mouse ovarian surface epithelial cell line (ID8 the efficacy of the bioluminescent signaling associated to

14
PISANO et al.

F I G U R E 5 Characterization of immune cell populations in ascites. (A) Representative image of ascites formation in tumor-bearing

populations in ascites collected from tumor-bearing mice (n = 3). (C) Immune cell populations identified from the mass cytometry analysis as
C57BL/6 mice 70 days after tumor cell injection. (B) visNE plot obtained by mass cytometry depicting the most represented immune cell

a percentage of the total cells. (D) Percentages of CD4+ and CD8+ T-cell subtypes, which are mainly CD4+/CD8+/IFNγ+,
CD4+/CD8+/PD1+, or CD8+/LyC6+/CD44+. Data presented are mean (SD)

ID8-Luc/GFP cells.34 Moreover, despite ID8 cells being aim of identifying a robust, reproducible protocol for
considered the gold standard when generating advanced tumor development. The range of concentrations selected
OC in immune competent mice,35 the scientific commu- was based on the most remarkable results found in lit-
nity is yet to provide robust protocols, nor a consensus erature, that is, significant tumor and acsities develop-
on optimal cell concentrations and incubation times for ment. Our results demonstrated significant tumor growth
tumor development. In particular, the literature reports a over a 9-week period for each of the cell concentrations

injected (between 1 × 106 and 1 × 107 ),32,36–39 and different

wide range of ID8 cell concentrations being peritoneally tested, suggesting the suitability of these cell concentra-
tions to reliably create a tumor in vivo. Additionally, the
incubation times required to develop a noticeable tumor onset of ascites was confirmed between 70 and 80 days after
and ascites in immunocompetent mice.34,39 cell injection, which was associated with increased mouse

Luc/GFP cells (5 × 106 , 1 × 107 , or 1.5 × 107 cells) for

In this work, we tested three concentrations of ID8- weight providing an indication of advanced stage disease.
H&E staining confirmed the presence of distinct tumor
their capacity to develop an advanced tumor in immune- nodules on the liver, scattered across the abdominal cav-
competent mice after intraperitoneal injection, with the ity and lining the peritoneal membrane. No differences in

15
PISANO et al.

F I G U R E 6 viSNE plots for 33 immune markers. From the CyTOF analysis of the 33 immune cell markers, three main immune cell
populations were identified based on the coexpression of specific markers. B cells: CD38+/B220+/MHC-II+/TNFα+/CD19+/CD80+. T cells
(CD4+ and CD8+): CD25+/PD-L1+/ PD-1+ have been colocalized on both, CD4+ and CD8+ cells, which also express subgroup-specific
markers TCRβ, IFN-γ, CD4, CD8α, Tbet, and CD103. The third subgroup is represented by neutrophils, eosinophils, macrophages, monocytes,
and dendritic cells, which are specifically positive for NK1.1, CD11c, Ly6G, SiglecF, and FoxP3. The markers CD86, F480, CD11b, and CD64 are
shared with the B cells subgroup, whereas RORgtm Ly6C, TCRgt, and Arg-1 are shared with the CD4+/CD8+ cell subgroup

16
PISANO et al.

TA B L E 2 Expression level of markers identified in each immune population found in the ascitic fluid

CD8+ CD4+ T NK Dendritic

B cells cells cells regulatory cells Monocytes Neutrophils Eosinophils Macrophages cells
CD38 ++ − − + − + + − + +
B220 ++ − − − − − − − − −
CD25 + + − ++ − − − − − −
TCRb − ++ ++ ++ − − − − − −
IFN-γ + ++ ++ − − − − − − +
MHC-II ++ − − + − − − − + +
TNF-α + − − − − − − − − −
PD-L1 ++ + + + − − − − + +
CD4 − − ++ ++ − − − − − −
Tbet − + − − + − − − − ++
CD19 ++ − − − − − − − − −
CD80 ++ − − − − + + + + −
PD-1 + + + + − − − − − −
CD8a − ++ − − − − − − − −
CD103 − + − + − − − − − −
CD86 + − − + − + + + − −
F480 − − − − − + + + ++ −
CD45 − − − − − − − − − −
RORgt − + − − − − − + − −
Ly6C − + − − − ++ ++ ++ − −
CD11b + − − − − + ++ ++ + ++
CD64 − − − − − + − − + −
CD44 ++ ++ + ++ ++ ++ ++ ++ ++ ++
TCRgt − − − − − + + + + −
Arg-1 − − − − − − − − − −
NK 1.1 − − − − ++ − − − − −
CD11c − − − − + − − − − ++
Ly6G − − − − − − ++ ++ − −
SinglecF − − − − − − − ++ − −
Foxp3 − − − ++ − − − − − −
iNOS − − − − − − − − − −
CD62L − − − − − − − − − −
CD206 − − − − − − − − − −

used to discern between high and low marker expression are reported in Section 2.6 of methods. ++: high expression, +: intermediate expression, –: lack of
The immune markers were assigned to each category of CD45+ cells according to their expression levels in that specific subpopulation. Marker intensity thresholds

expression. iNOS, CD62L, and CD206 were not expressed on any cell population within the ascites.

the number or size of nodules between the experimental abdomen. The presence of CD3+ tumor infiltrating lym-
groups was evident microscopically. Thus, further char- phocytes (TILs) has been identified as an independent
acterization to assess the immune environment within prognostic factor in patients with epithelial OC.41–45 In
the tumor nodules was undertaken. In agreement with addition, antigen presenting cells such as tumor associ-
the existing literature on immune cells present within ated macrophages (TAMs) and DCs have significant roles
the OC TME,40 immunohistochemical analysis showed T in the TME. In particular, TAMs, the most represented
cells and antigen presenting cells (macrophages and DCs) cell population,46 have the potential to suppress or stimu-
distributed throughout tumor nodules found on the sur- late an anticancer response according to the effect the sur-
face of the peritoneal membrane and scattered within the rounding microenvironment exerts on them.47,48

17
PISANO et al.

F I G U R E 7 Schematic of ascites immune cells and their interactions with the tumor. The interactions between ascites immune cells and
their potential effect on tumor cells. B220+/CD11b+/CD38+ B lymphocytes (A) and their CD25+/CD119+ regulatory B-cell subgroup (B)
exert antitumor and protumor activity, respectively. The role of CD4+ (C) and CD8+ (D) T cells in exerting cytotoxic activity towards the
tumor is linked to the antigen presenting capacity of CD11b+/CD11c+/CD44+ DCs (E). Eosinophils (G) and neutrophils (F) are linked to
protumoral and antitumoral properties, respectively. In parallel, monocytes and macrophages (H-J) exert either a tumor promoting or
suppressive effect according to the surrounding microenvironment

Focusing on the ascites, which is closely linked to an immunosuppressive climate that might lead to a decreased
altered immune environment within the peritoneal cavity activation of ascites-derived T cells.50
of advanced OC patients, further characterization was In Figure 7, we summarize the possible interactions
undertaken. To comprehensively analyze the immune occurring between the immune cell populations identified
landscape of the ascitic fluid collected from an ID8 within the ascites and their effect on the metastatic tumors
ovarian cancer model, for the first time we exploited in situ, based on cell-surface marker expression. The pres-
mass cytometry. The 33 immune cell markers analyzed ence of CD45+B220+ B cells can be linked to both pro-
allowed for the specific identification of distinct immune and antitumor responses due to their phenotypical and
cell populations and their linkage to pivotal functions functional variability, as confirmed by several other stud-
with respect to tumor progression. Compared to the ies. Indeed, the B220+CD11b+MHC-II+ B-cell population
review from Wertel et al mentioned above, our results (7A) can have a positive or neutral prognostic effect,51,52
showed similar overall percentages of CD8+ T cells, in which can also be mediated by CD38 expression.53 In con-
contrast with the lower percentages of CD4+ T cells and B trast to this, the presence of CD25+CD19+ B regulatory
cells.49 The CD8+/CD4+ T-cell ratio of 1.76 we identified cells (7B) is correlated with suppressed T cells responses
was indicative of a higher overall presence of CD8+ T and poorer patient survival.54
cells compared to CD4+ cells, which is associated with Two additional markers, PD-1 and PD-L1, present on
improved patient survival. This finding, together with the both CD4+ (7C) and CD8+ T (7D) cells, have a pivotal role
expression of IFNγ, suggests effective immune stimulation in establishing efficient immunotherapeutic approaches,
within the tumor and the development of cell-mediated after it was demonstrated that their inhibition can stop
immunity. In contrast, Giuntoli et al demonstrated that cancer progression.55 Although clinical trials testing PD-1
a high CD4+/CD8+ T-cell ratio is associated with poor and PD-L1 inhibitors have not yet yielded satisfactory
outcome done in patients with ovarian, primary peritoneal results in OC as single treatment,8,9 their use as combina-
or fallopian tube cancers, and that high concentrations of torial treatment still holds promise. The prognostic value
interleukins 6 (IL-6) and 10 (IL-10) can help establish an of PD1+ TILs, when colocalized with PD-L1 on cancer cells

18
PISANO et al.

has been demonstrated supporting the PD-1 inhibitory indicate a better prognosis,64–66 the high expression of
pathway as one mechanism they use to silence the Ly6C on monocytes is a strong indicator of a TAM pheno-
immune system during OC progression.56 In particular, type with strong immunosuppressive potential and poor a
although tumors appear to be infiltrated by T cells at early prognosis.67,68
stages, a progressive reduction in the frequency of CD8+ More generally, the presence of immune-active compo-
T cells and CD8:Treg ratio was noticed at more advanced nents within the tumor nodules and the ascites raises the
advanced stages.57 Our findings have also attributed the question of why the therapeutic potential of immunother-
majority of PD-L1 expression to macrophages (7H) and, apies is still limited in OC settings. In this case, addi-
together with the presence of cytolytic and regulatory TIL tional factors should be considered, including the so-called
subsets, link directly to survival potential.58 Macrophages, tumor mutation burden (TMB). TMB results from the iden-
identified through the coexpression of F4/80 and CD64, tification and quantification of driver genes mutations that
were also positive for MHC-II, CD44, and CD80, with the are responsible for the production of neoantigens. The
latter marker suggesting an M1 phenotype, which has been increasing presence of neoantigens has been associated to
linked to increased inflammatory status47 and is specif- the activation of the antitumor immune response. For this
ically correlated with a longer overall survival (OS) and reason, TBM plays an important role in the progression of
progression-free survival (PFS) in serous OC patients.59 a cancer with a high mutation load being associate to a
Neutrophils (7F), identified by the presence of the better prognosis.69 A recent investigation calculating TMB
markers Ly6C and Ly6G (similar to other myeloid in 397 patients with OC in the TCGA database revealed
derived populations such as eosinophils (7G) and mono- that resting immune cells (B cells, B cells, CD4+ T cells,
cytes/macrophages), were also present.60 Neutrophils Tregs, monocytes, mast cells, and neutrophils) likely infil-
have been connected to antitumor-promoting activity trate tumors with low TMB, whereas activated immune
in OC. Indeed, neutrophils isolated from the ascites cells (CD4+ T cells, follicle-assisted T cells, proinflamma-
of a KRAS-ID8-induced mouse model showed KRAS- tory macrophages) infiltrate tumors with high TMB.70 In
dependent CD8+ T-cell activation through increased other cases, some cell-based immunotherapies (such as
recruitment of costimulatory molecules. On the con- CAR-T) targeting a single tumor antigen often lose their
trary, neutrophil depletion (through administration of an efficacy as the result of mutations occurring in tumor
anti-Ly6G monoclonal antibody) led to marked tumor cells, which impair specific antigen expression thus hin-
progression.37 More recently, however, Ly6G-positive neu- dering the effect of the therapy.71 In addition, cell therapeu-
trophils have been reported to promote a microenviroment tics often are subjected to the immunosuppressive envi-
that is conducive of metastases spreading and accumula- ronment they meet following administration, which lim-
tion at specific sites.61 its their effectiveness in exerting an antitumor immune
Dendritic cells, identified by the expression of CD11b+, response.28,72 The tryptophan catabolism offers another
CD11c+, CD44+, and MHC-II (7E), are paramount players example relevant in this context, as the tryptophan-
in the activation of effective T-cell responses through their catabolizing enzyme indoleamine 2,3-dioxygenase (IDO)
antigen-presenting activity. Indeed, CD44 was found to be has been found to be hyperactive in OC and linked to
pivotal in the formation of tight junctions between mature the production of immunosuppressive catabolites and poor
DCs and T cells and to play a role in T-cell activation patient survival.73 In addition, the cancer-induced acidic
as a consequence.62 However, DCs can undergo tumor- environment has been shown to have a role in tumor recur-
mediated immunosupressive processes, such as the block- rence, metastasis, and prognosis of cancer patients (due
age of their activity through the tumor-induced upregula- to the high production of lactate).74 Furthermore, lactate
tion of the unfolded protein response (UPR), as showed can also support cancer cell immune evasion by inhibit-
by Cubillos-Ruiz et al.14 Moreover, Krempski et al also ing T-cell activation75 and dendritic cell antigen presenting
found that tumor infiltrating, PD-1+/PD-L1+ DCs within capacities.76
the ascites respond poorly to danger signal, suppress T-cell This work is the first to provide a multiparametric
activity and decrease T-cell infiltration within the tumor and comprehensive characterization of the immune cell
masses.63 landscape of the ascites collected from a preclinical model
The immune cells identified within the ascites produced of advanced OC. Published literature reports fragmented
in this model of HGSOC are linked to both pro- and antitu- information, as only single populations (such as CD4+ and
moral activity, indicating that this model represents a bal- CD8+ T cells) have been so far identified and described.50
anced immune response to the tumor, or that the immuno- A more complete description has been offered by Wertel
suppressive effect of the tumor is yet to take hold. For et al who listed the percentages of the main cellular
instance, while expression of the integrin, CD103, and tran- components found in the peritoneal fluid of advanced OC
scription factor, Tbet, associated with CD8+ T cells might patients by merging the information collected from several

19
PISANO et al.

different studies.49 More recently, the panorama of the (WEFO) under the European Regional Development Fund
ascites collected from HGSOC patients has been resolved (ERDF) and Houston Methodist Research Institute. SP
by applying single cell-RNA sequencing (scRNA-seq).77 is sponsored by the Swansea University Medical School
In this study, the authors provided a broad view of the (UK)/Houston Methodist Research Institute (US) joint
different cell types in the ascites ecosystem, with particular PhD Initiative. Additional support for the study was pro-
focus on malignant versus nonmalignant cells (analyzing vided by the Golfers Against Cancer Foundation.
samples partially depleted of CD45+ immune cells).
The potential strength of the data we identified is there- COMPETING INTEREST
fore to demonstrate that mass cytometry provides a plat- The authors declare that they have no competing interests.
form for the comprehensive analysis of the immune cell
landscape within ascites, which would allow periodical E T H I C S A P P R O VA L A N D C O N S E N T T O
analysis of cellular and molecular changes in patients PA R T I C I PA T E
with OC. In this regard, CyTOF holds the promise of All animal studies were carried out in accordance with
complementing personalized therapeutic approaches, and guidelines determined by the Animal Welfare Act and the
potentially enabling real time tracking of the efficacy of Guide for the Care and Use of Laboratory Animals and
immunotherapeutics. Compared to scRNA-seq, CYTOF complied with protocols approved by the Institutional Ani-
offers the advantage of a higher throughput for the evalua- mal Care and Use Committee at the Houston Methodist
tion of the TME in clinical samples, as it allows for a more Research Institute (AUP-0219-0013).

of >30 selected antigen markers. Moreover, CyTOF “nar-

accurate targeting of immune cell subsets through the use
AU T H O R S ’ CO N T R I B U T I O N S
row and distinct”78,79 data are generated from the anal- Conceptualization and methodology: SP, BC; Formal anal-
ysis of over 250 000 cells, whereas transcriptome-based ysis and data curation: SP, SL, FI, LM, OSV, GDH, RSC and
platforms detect wider unbiased populations from several BC; Validation and investigation: SP, SL, LM and BC; Orig-
thousands of cells.80,81 inal draft preparation and Writing: SP, LM, GDH and BC;
Review and editing: SP, GDH, DG, RSC and BC; Approval
of final manuscript: all authors read and approved the final
5 CONCLUSIONS manuscript.

In this work, we provide, for the first time, a compre- AVA I L A B I L I T Y O F D A T A A N D

hensive characterization of the immune landscape of the M AT E R I A L
ascites collected from tumor-bearing mice, unveiling its All data relevant to the study are included in the article or
potential for clinical implementation. The continuous uploaded as supplementary information.
analysis of interactions between immune cells in a cancer-
ous environment would significantly increase the number ORCID
of therapeutic options for the treatment of this malignancy Simone Pisano https://orcid.org/0000-0002-5412-1241
and offer a significant alternative for the evaluation of Yajaira S. Jimenez https://orcid.org/0000-0003-4647-
ongoing therapies. Data presented in this study prove mass 4546
cytometry as a promising tool to facilitate this process,
with the potential to identify personalized therapeutic
REFERENCES
targets and establish improved immunotherapy strategies.
1. Siegel RL, Miller KD, Jemal A. Cancer statistics, 2020. CA Can-
In addition, the application of CyTOF on more complex
cer J Clin. https://doi.org/10.3322/caac.21590. Published Online
and genetically modified mice models, as well as on First: 2020.
patients’ derived samples, will also unveil new insights 2. Wu J, Sun H, Yang L, et al. Improved survival in ovarian cancer,
into disease heterogeneity, pathology, and drug resistance, with widening survival gaps of races and socioeconomic status:
and will expand our understanding of HGSOC. a period analysis. J Cancer. 2018;9:3548-3556. https://doi.org/10.
7150/jca.26300.
AC K N OW L E D G M E N T 3. Torre LA, Trabert B, DeSantis CE, et al. Ovarian cancer statis-
tics, 2018. CA Cancer J Clin. https://doi.org/10.3322/caac.21456.
We thank the Houston Methodist Research Institute
Published Online First: 2018.
(HMRI) Immunomonitoring Core. BC acknowledges sup-
4. Lisio MA, Fu L, Goyeneche A, et al. High-grade serous ovarian
port through the Sêr Cymru II scheme, funded by the cancer: basic sciences, clinical and therapeutic standpoints. Int
European Union’s Horizon 2020 Research and Innovation J Mol Sci. 2019;20(4):952. https://doi.org/10.3390/ijms20040952.
Program under the Marie Skłodowska-Curie grant agree- 5. Armstrong DK, Alvarez RD, Bakkum-Gamez JN, et al. NCCN
ment No. 663830, the Welsh European Funding Office Guidelines Insights: Ovarian Cancer, Version 1.2019. J Natl

20
PISANO et al.

Compr Cancer Netw. 2019;17(8):896-909. https://doi.org/10.6004/ Immunother. 2001;24:392-407. https://doi.org/10.1097/

jnccn.2019.0039. 00002371-200109000-00002.
6. Garcia A, Singh H. Bevacizumab and ovarian cancer. Ther 22. Moser M. Dendritic cells in immunity and tolerance—do they
Adv Med Oncol. 2013;5(2):133-141. https://doi.org/10.1177/ display opposite functions? Immunity. 2003;19:5-8. https://doi.
1758834012467661. org/10.1016/S1074-7613(03)00182-1.
7. Vasey PA. Resistance to chemotherapy in advanced ovar- 23. Nelson BH. The impact of T-cell immunity on ovarian can-
ian cancer: mechanisms and current strategies. Br J Can- cer outcomes. Immunol Rev. 2008;222:101-116. https://doi.org/10.
cer. 2003;89:S23–S28. https://doi.org/10.1038/sj.bjc.6601497. Pub- 1111/j.1600-065X.2008.00614.x.
lished Online First: 2003. 24. Toker A, Nguyen LT, Stone SC, et al. Regulatory T cells in ovar-
8. Matulonis UA, Shapira-Frommer R, Santin AD, et al. Antitumor ian cancer are characterized by a highly activated phenotype dis-
activity and safety of pembrolizumab in patients with advanced tinct from that in melanoma. Clin Cancer Res. 2018;24(22):5685–
recurrent ovarian cancer: results from the phase II KEYNOTE- 5696. https://doi.org/10.1158/1078-0432.CCR-18-0554. Published
100 study. Ann Oncol. https://doi.org/10.1093/annonc/mdz135. Online First: 2018.
Published Online First: 2019. 25. Toyoshimaa M, Tanaka Y, Matumoto M, et al. Generation of a
9. Bartl T, Paspalj V, Polterauer S, et al. Current state and perspec- syngeneic mouse model to study the intraperitoneal dissemina-
tives of checkpoint inhibitors in ovarian cancer treatment. Memo tion of ovarian cancer with in vivo luciferase imaging. Lumines-
- Mag Eur Med Oncol. 2020;13:202-206. https://doi.org/10.1007/ cence. https://doi.org/10.1002/bio.1112. Published Online First:
s12254-020-00579-z. 2009.
10. Galon J, Bruni D. Approaches to treat immune hot, altered 26. Fienberg HG, Simonds EF, Fantl WJ, et al. A platinum-
and cold tumours with combination immunotherapies. Nat based covalent viability reagent for single-cell mass cytometry.
Rev Drug Discov. 2019;18:197-218. https://doi.org/10.1038/ Cytom Part A. 2012;81 A:467-475. https://doi.org/10.1002/cyto.a.
s41573-018-0007-y. 22067.
11. Ghisoni E, Imbimbo M, Zimmermann S, et al. Ovarian can- 27. Pokhriyal R, Hariprasad R, Kumar L, et al. Chemotherapy
cer immunotherapy: turning up the heat. Int J Mol Sci. resistance in advanced ovarian cancer patients. Biomark
2019;20(12):2927. https://doi.org/10.3390/ijms20122927. Cancer. 2019;11:1179299X19860815. https://doi.org/10.1177/
12. Lanitis E, Dangaj D, Irving M, et al. Mechanisms regulating 1179299X19860815.
T-cell infiltration and activity in solid tumors. Ann Oncol. 28. Nizzero S, Shen H, Ferrari M, et al. Immunotherapeutic trans-
2017;28(suppl_12):xii18-xii32. https://doi.org/10.1093/annonc/ port oncophysics: space, time, and immune activation in cancer.
mdx238. Trends Cancer. 2020;6(1):40-48. https://doi.org/10.1016/j.trecan.
13. Scarlett UK, Rutkowski MR, Rauwerdink AM, et al. Ovarian can- 2019.11.008.
cer progression is controlled by phenotypic changes in dendritic 29. Roby KF, Taylor CC, Sweetwood JP, et al. Development of
cells. J Exp Med. https://doi.org/10.1084/jem.20111413. Published a syngeneic mouse model for events related to ovarian can-
Online First: 2012. cer. Carcinogenesis. 2000;21(4):585-591. https://doi.org/10.1093/
14. Cubillos-Ruiz JR, Silberman PC, Rutkowski MR, et al. ER stress carcin/21.4.585.
sensor XBP1 controls anti-tumor immunity by disrupting den- 30. Maniati E, Berlato C, Gopinathan G, et al. Mouse Ovarian Can-
dritic cell homeostasis. Cell. 2015;161(7):1527–1538. https://doi. cer Models Recapitulate the Human Tumor Microenvironment
org/10.1016/j.cell.2015.05.025. Published Online First: 2015. and Patient Response to Treatment. Cell Rep. 2020;30:525-540.e7.
15. Roane BM, Arend RC, Birrer MJ. Review: targeting the trans- https://doi.org/10.1016/j.celrep.2019.12.034.
forming growth factor-beta pathway in ovarian cancer. Cancers 31. Soriano AA, De Cristofaro T, Palma Di, et al. PAX8 expression
(Basel). 2019;11(5):668. https://doi.org/10.3390/cancers11050668. in high-grade serous ovarian cancer positively regulates attach-
16. Kao JY, Gong Y, Chen C-M, et al. Tumor-derived TGF-β reduces ment to ECM via Integrin β3. Cancer Cell Int. 2019;19(303):1-12.
the efficacy of dendritic cell/tumor fusion vaccine. J Immunol. https://doi.org/10.1186/s12935-019-1022-8.
https://doi.org/10.4049/jimmunol.170.7.3806. Published Online 32. Wilson AJ, Barham W, Saskowski J, et al. Tracking NF-κB activ-
First: 2003. ity in tumor cells during ovarian cancer progression in a syn-
17. Ahmed N, Stenvers KL. Getting to know ovarian cancer ascites: geneic mouse model. J Ovarian Res. 2013;6(1):63. https://doi.org/
opportunities for targeted therapy-based translational research. 10.1186/1757-2215-6-63.
Front Oncol. 2013;3:256. https://doi.org/10.3389/fonc.2013.00256. 33. Zhang L, Yang N, Conejo Garcia JR, et al. Generation of a syn-
18. Kipps E, Tan DSP, Kaye SB. Meeting the challenge of ascites geneic mouse model to study the effects of vascular endothelial
in ovarian cancer: new avenues for therapy and research. growth factor in ovarian carcinoma. Am J Pathol. 2002;161:2295-
Nat Rev Cancer. 2013;13(4):273-282. https://doi.org/10.1038/ 2309. https://doi.org/10.1016/S0002-9440(10)64505-1.
nrc3432. 34. Baert T, Verschuere T, Van Hoylandt A, et al. The dark side of
19. Cohen M, Petignat P. The bright side of ascites in ovarian cancer. ID8-Luc2: pitfalls for luciferase tagged murine models for ovar-
Cell Cycle. 2014;13(15):2319. https://doi.org/10.4161/cc.29951. ian cancer. J Immunother Cancer. 2015;3:57. https://doi.org/10.
20. Tan DSP, Agarwal R, Kaye SB. Mechanisms of transcoelomic 1186/s40425-015-0102-0.
metastasis in ovarian cancer. Lancet Oncol. 2006;7(11):925-934. 35. Gil M, Komorowski MP, Seshadri M, et al. CXCL12/CXCR4
https://doi.org/10.1016/S1470-2045(06)70939-1. blockade by oncolytic virotherapy inhibits ovarian cancer
21. Mocellin S, Wang E, Marincola FM. Cytokines and growth by decreasing immunosuppression and targeting cancer-
immune response in the tumor microenvironment. J initiating cells. J Immunol. 2014;193:5327-5337. https://doi.org/
10.4049/jimmunol.1400201.

21
PISANO et al.

36. Cho S, Sun Y, Soisson AP, et al. Characterization and evaluation for antitumor immunity. Anticancer Res. 2009;29:2875-2884. http:
of pre-clinical suitability of a syngeneic orthotopic mouse ovar- //www.ncbi.nlm.nih.gov/pubmed/19661290.
ian cancer model. Anticancer Res. 2013;33(4):1317-1324. 51. Wouters MCA, Nelson BH. Prognostic significance of tumor-
37. Yoshida M, Taguchi A, Kawana K, et al. Intraperitoneal infiltrating B cells and plasma cells in human cancer. Clin
neutrophils activated by KRAS-induced ovarian cancer exert Cancer Res. 2018;24:6125-6135. https://doi.org/10.1158/1078-0432.
antitumor effects by modulating adaptive immunity. Int J Oncol. CCR-18-1481.
2018;53:1580-1590. https://doi.org/10.3892/ijo.2018.4504. 52. Gupta P, Chen C, Chaluvally-Raghavan P, et al. B cells as an
38. Zhu X, Xu J, Cai H, et al. Carboplatin and programmed death- immune-regulatory signature in ovarian cancer. Cancers (Basel).
ligand 1 blockade synergistically produce a similar antitumor 2019;11(7):894. https://doi.org/10.3390/cancers11070894.
effect to carboplatin alone in murine ID8 ovarian cancer model. J 53. Zhu Y, Zhang Z, Jiang Z, et al. CD38 predicts favorable progno-
Obstet Gynaecol Res. 2018;44:303-311. https://doi.org/10.1111/jog. sis by enhancing immune infiltration and antitumor immunity
13521. in the epithelial ovarian cancer microenvironment. Front Genet.
39. Liao JB, Ovenell KJ, Curtis EEM, et al. Preservation of tumor- 2020;11. https://doi.org/10.3389/fgene.2020.00369.
host immune interactions with luciferase-tagged imaging in 54. Wei X, Jin Y, Tian Y, et al. Regulatory B cells con-
a murine model of ovarian cancer. J Immunother Cancer. tribute to the impaired antitumor immunity in ovar-
2015;25(3):16. https://doi.org/10.1186/s40425-015-0060-6. ian cancer patients. Tumor Biol. 2016;37(5):6581-6588.
40. Cai DL, Jin LP. Immune cell population in ovarian tumor https://doi.org/10.1007/s13277-015-4538-0.
microenvironment. J Cancer. 2017;8(15):2915-2923. https://doi. 55. Gong J, Chehrazi-Raffle A, Reddi S, et al. Development of PD-
org/10.7150/jca.20314. 1 and PD-L1 inhibitors as a form of cancer immunotherapy: a
41. Tomšová M, Melichar B, Sedláková I, et al. Prognostic signifi- comprehensive review of registration trials and future considera-
cance of CD3+ tumor-infiltrating lymphocytes in ovarian carci- tions. J Immunother Cancer. 2018;6(1):8. https://doi.org/10.1186/
noma. Gynecol Oncol. 2008;108:415-420. https://doi.org/10.1016/ s40425-018-0316-z.
j.ygyno.2007.10.016. 56. Wang Q, Lou W, Di W, et al. Prognostic value of tumor PD-L1
42. Stumpf M, Hasenburg A, Riener MO, et al. Intraepithelial CD8- expression combined with CD8+ tumor infiltrating lympho-
positive T lymphocytes predict survival for patients with serous cytes in high grade serous ovarian cancer. Int Immunophar-
stage III ovarian carcinomas: relevance of clonal selection of T macol. 2017;52:7-14. https://doi.org/10.1016/j.intimp.2017.08.
lymphocytes. Br J Cancer. 2009;101:1513-1521. https://doi.org/10. 017.
1038/sj.bjc.6605274. 57. Duraiswamy J, Freeman GJ, Coukos G. Therapeutic PD-1 path-
43. Raspollini MR, Castiglione F, Degl’Innocenti DR, et al. Tumour- way blockade augments with other modalities of immunother-
infiltrating gamma/delta T-lymphocytes are correlated with apy T-cell function to prevent immune decline in ovarian can-
a brief disease-free interval in advanced ovarian serous car- cer. Cancer Res. 2013;73(23):6900–6912. https://doi.org/10.1158/
cinoma. Ann Oncol. 2005;16:590-596. https://doi.org/10.1093/ 0008-5472.CAN-13-1550. Published Online First: 2013.
annonc/mdi112. 58. Webb JR, Milne K, Kroeger DR, et al. PD-L1 expression is asso-
44. Leffers N, Gooden MJM, De Jong RA, et al. Prognostic sig- ciated with tumor-infiltrating T cells and favorable prognosis in
nificance of tumor-infiltrating T-lymphocytes in primary and high-grade serous ovarian cancer. Gynecol Oncol. 2016;141:293-
metastatic lesions of advanced stage ovarian cancer. Cancer 302. https://doi.org/10.1016/j.ygyno.2016.03.008.
Immunol Immunother. 2009;58(3):449-459. https://doi.org/10. 59. Macciò A, Gramignano G, Cherchi MC, et al. Role of M1-
1007/s00262-008-0583-5. polarized tumorassociated macrophages in the prognosis of
45. Zhang L, Conejo-Garcia JR, Katsaros D, et al. Intratumoral advanced ovarian cancer patients. Sci Rep. 2020;10. https://doi.
T cells, recurrence, and survival in epithelial ovarian can- org/10.1038/s41598-020-63276-1.
cer. N Engl J Med. 2003;348:203-213. https://doi.org/10.1056/ 60. Rose S, Misharin A, Perlman H. A novel Ly6C/Ly6G-based strat-
NEJMoa020177. egy to analyze the mouse splenic myeloid compartment. Cytom
46. Drakes ML, Stiff PJ. Regulation of ovarian cancer prognosis by Part A. 2012;81 A:343-350. https://doi.org/10.1002/cyto.a.22012.
immune cells in the tumor microenvironment. Cancers (Basel). 61. Lee WJ, Ko SY, Mohamed MS, et al. Neutrophils facilitate ovar-
2018;10(9):302. https://doi.org/10.3390/cancers10090302. ian cancer premetastatic niche formation in the omentum. J Exp
47. Allavena P, Sica A, Garlanda C, et al. The Yin-Yang of tumor- Med. 2019;216:176-194. https://doi.org/10.1084/jem.20181170.
associated macrophages in neoplastic progression and immune 62. Hegde VL, Singh NP, Nagarkatti PS, et al. CD44 mobilization
surveillance. Immunol Rev. 2008;222:155-161. https://doi.org/10. in allogeneic dendritic cell-T cell immunological synapse plays
1111/j.1600-065X.2008.00607.x. a key role in T cell activation. J Leukoc Biol. https://doi.org/10.
48. Zhang M, He Y, Sun X, et al. A high M1/M2 ratio of tumor- 1189/jlb.1107752. Published Online First: 2008.

infiltrating programmed death receptor-1 + dendritic cells

associated macrophages is associated with extended survival in 63. Krempski J, Karyampudi L, Behrens MD, et al. Tumor-
ovarian cancer patients. J Ovarian Res. 2014;7:19. https://doi.org/
10.1186/1757-2215-7-19. mediate immune suppression in ovarian cancer. J Immunol.
49. Wertel I, Nowicka A, Rogala E, et al. Peritoneal immune sys- 2011;186:6905-6913. https://doi.org/10.4049/jimmunol.1100274.
tem in patients with advance epithelial ovarian cancer. Int Rev 64. Webb JR, Milne K, Watson P, et al. Tumor-infiltrating lympho-
Immunol. 2011;30:87-101. https://doi.org/10.3109/08830185.2011. cytes expressing the tissue resident memory marker cd103 are
569902. associated with increased survival in high-grade serous ovarian
50. Giuntoli RL, Webb TJ, Zoso A, et al. Ovarian cancer-associated cancer. Clin Cancer Res. 2014;20(2):434-444. https://doi.org/10.
ascites demonstrates altered immune environment: implications 1158/1078-0432.CCR-13-1877.

22
PISANO et al.

65. Xu Y, Chen L, Xu B, et al. Higher numbers of T-bet + tumor- 2007;109(9):3812-3819. https://doi.org/10.1182/blood-2006-07-

infiltrating lymphocytes associate with better survival in human 035972.
epithelial ovarian cancer. Cell Physiol Biochem. 2017;41:475-483. 76. Gottfried E, Kunz-Schughart LA, Ebner S, et al. Tumor-derived
https://doi.org/10.1159/000456600. lactic acid modulates dendritic cell activation and antigen
66. Gacerez AT, Sentman CL. T-bet promotes potent antitumor expression. Blood. 2006;107:2013-2021. https://doi.org/10.1182/
activity of CD4+ CAR T cells. Cancer Gene Ther. 2018;25(5-6):117- blood-2005-05-1795.
128. https://doi.org/10.1038/s41417-018-0012-7. 77. Izar B, Tirosh I, Stover EH, et al. A single-cell landscape of high-
67. Movahedi K, Laoui D, Gysemans C, et al. Different tumor grade serous ovarian cancer. Nat Med. 2020;26:1271-1279. https:
microenvironments contain functionally distinct subsets of //doi.org/10.1038/s41591-020-0926-0.
macrophages derived from Ly6C(high) monocytes. Cancer 78. Kashima Y, Togashi Y, Fukuoka S, et al. Potentiality of mul-
Res. 2010;70(14):5728-5739. https://doi.org/10.1158/0008-5472. tiple modalities for single-cell analyses to evaluate the tumor
CAN-09-4672. microenvironment in clinical specimens. Sci Rep. 2021;11:341.
68. Reinartz S, Schumann T, Finkernagel F, et al. Mixed- https://doi.org/10.1038/s41598-020-79385-w.
polarization phenotype of ascites-associated macrophages 79. Schelker M, Feau S, Du J, et al. Estimation of immune cell con-
in human ovarian carcinoma: correlation of CD163 expression, tent in tumour tissue using single-cell RNA-seq data. Nat Com-
cytokine levels and early relapse. Int J Cancer. 2014;134:32-42. mun. 2017;8:2032. https://doi.org/10.1038/s41467-017-02289-3.
https://doi.org/10.1002/ijc.28335. 80. Bacher R, Kendziorski C. Design and computational analy-
69. Snyder A, Makarov V, Merghoub T, et al. Genetic basis for clin- sis of single-cell RNA-sequencing experiments. Genome Biol.
ical response to CTLA-4 blockade in melanoma. N Engl J Med. 2016;17(63). https://doi.org/10.1186/s13059-016-0927-y.
2014;371:2189-2199. https://doi.org/10.1056/nejmoa1406498. 81. Ren X, Kang B, Zhang Z. Understanding tumor ecosystems by
70. Bi F, Chen Y, Yang Q. Significance of tumor mutation burden single-cell sequencing: promises and limitations 11 medical and
combined with immune infiltrates in the progression and prog- health sciences 1112 oncology and carcinogenesis 06 biological
nosis of ovarian cancer. Cancer Cell Int. 2020;20:373. https://doi. sciences 0604 genetics. Genome Biol. 2018;19(211). https://doi.
org/10.1186/s12935-020-01472-9. org/10.1186/s13059-018-1593-z.
71. Mehta A, Kim YJ, Robert L, et al. Immunotherapy resis-
tance by inflammation-induced dedifferentiation. Cancer Dis-
cov. 2018;8(8):935-943. https://doi.org/10.1158/2159-8290.CD-17- S U P P O RT I N G I N F O R M AT I O N
1178. Additional supporting information may be found online
72. Corradetti B, Pisano S, Conlan RS, et al. Nanotechnology and
in the Supporting Information section at the end of the
immunotherapy in ovarian cancer: tracing new landscapes.
J Pharmacol Exp Ther. 2019;370(3):636-646. https://doi.org/10.
article.
1124/jpet.118.254979.
73. Smith LP, Bitler BG, Richer JK, et al. Tryptophan catabolism
in epithelial ovarian carcinoma. Trends Cancer Res. 2019;14:1-9.
http://www.ncbi.nlm.nih.gov/pubmed/31736606.
74. Kato Y, Ozawa S, Miyamoto C, et al. Acidic extracellular
microenvironment and cancer. Cancer Cell Int. 2013;13(89):1-8.
https://doi.org/10.1186/1475-2867-13-89.
75. Fischer K, Hoffmann P, Voelkl S, et al. Inhibitory effect
of tumor cell-derived lactic acid on human T cells. Blood.

23
DOI: 10.1002/eji.202149329 Eur. J. Immunol. 2022. 52: 96–108
Clinical

Systems immunology

Research Article
Systematic analysis of CD39, CD103, CD137, and PD-1
as biomarkers for naturally occurring tumor
antigen-specific TILs
Monika A. Eiva1,2,3 , Dalia K. Omran1 , Jessica A. Chacon4
and Daniel J. Powell Jr.1,2,3

1
Ovarian Cancer Research Center, Department of Obstetrics and Gynecology, Perelman School
of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA
2
Center for Cellular Immunotherapies, Abramson Cancer Center, University of Pennsylvania,
Philadelphia, Pennsylvania, USA
3
Department of Pathology and Laboratory Medicine, Abramson Cancer Center, Perelman
School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA
4
Paul L Foster School of Medicine and Woody L. Hunt School of Dental Medicine, Texas Tech
University Health Sciences Center, El Paso, Texas, USA

The detection of tumor-specific T cells in solid tumors is integral to interrogate endoge-

nous antitumor responses and to advance downstream therapeutic applications. Mul-
tiple biomarkers are reported to identify endogenous tumor-specific tumor-infiltrating
lymphocytes (TILs), namely CD137, PD-1, CD103, and CD39; however, a direct compar-
ison of these molecules has yet to be performed. We evaluated these biomarkers in pri-
mary human ovarian tumor samples using single-cell mass cytometry to compare their
relative phenotypic profiles, and examined their response to autologous tumor cells ex
vivo. PD-1+ , CD103+ , and CD39+ TILs all contain a CD137+ cell subset, while CD137+ TILs
highly co-express the aforementioned markers. CD137+ TILs exhibit the highest expres-
sion of cytotoxic effector molecules compared to PD-1+ , CD103+ , or CD39+ TILs. Removal
of CD137+ cells from PD-1+ , CD103+ , or CD39+ TILs diminish their IFN-γ secretion in
response to autologous tumor cell stimulation, while CD137+ TILs maintain high HLA-
dependent IFN-γ secretion. CD137+ TILs exhibited an exhausted phenotype but with CD28
co-expression, suggesting possible receptiveness to reinvigoration via immune check-
point blockade. Together, our findings demonstrate that the antitumor abilities of PD-1+ ,
CD103+ , and CD39+ TILs are mainly derived from a subset of CD137-expressing TILs,
implicating CD137 as a more selective biomarker for naturally occurring tumor-specific
TILs.

Keywords: CD39 CD103 CD137 PD-1 tumor-infiltrating lymphocytes

Additional supporting information may be found online in the Supporting Information section
at the end of the article.

Correspondence: Daniel J. Powell Jr.

e-mail: [email protected]

© 2021 Wiley-VCH GmbH www.eji-journal.eu

24
Eur. J. Immunol. 2022. 52: 96–108 Systems immunology

Introduction Results

The intratumoral abundance of tumor-infiltrating lymphocytes A subset of TILs express effector molecules
(TILs) is a positive prognostic factor for increased survival in most
solid cancers, indicating that TILs are integral to endogenous anti- To investigate the phenotype of TILs harbored within infiltrated
tumor immunity and play a role in controlling cancer progression tumors, the algorithms viSNE and PhenoGraph metaclustering
[1, 2]. However, only a small percentage of TILs respond against [16, 17] were used to co-map CD3+ CD45+ TILs in enzyme-
tumor antigens and their antitumor response can be hindered digested ovarian tumors analyzed by single-cell mass cytome-
by multiple mechanisms of immunosuppression [3, 4]. The chal- try. To address patient-specific variability and to understand TIL
lenges of detecting TILs capable of responding to tumor antigens dynamics shared between samples, PhenoGraph clusters were
have led to great interest in identifying biomarkers of tumor- merged using the metaclustering algorithm in the interactive cyt
specific TILs in solid tumors. Biomarkers that identify tumor- tool [16]. Metaclustering analysis identified seven major TIL pop-
specific TILs are integral for downstream applications, such as ulations (Fig. 1A). Metaclusters (MCs) 1, 5, 6, and 7 were gener-
enriching tumor-specific TILs for use in adoptive cellular therapy, ally conserved among all samples tested, while MCs 2, 3, and 4
investigating endogenous antitumor immunity, studying mech- had greater variability (Fig. 1B). MC5 (mean = 1.51, 95% CI =
anisms of effective immunotherapy, identifying antigen-specific -0.37 to 3.39) and MC6 (mean = 2.21, 95% CI = -0.83 to 5.24)
T-cell receptors or neoantigens, and exploring the immunobiology were the rarest subsets in all samples, and MC5 was consistently
of these cells [5–8]. The need for effective biomarkers to detect enriched for cells expressing activation, proliferation, and effec-
T cells is further underscored by the fact that many cancers, such tor molecules (Fig. 1B,C). Compared to the other metaclustered
as ovarian cancer, do not have well-defined shared tumor-specific groups, only MC5 highly expressed effector molecules associated
antigens capable of initiating a tumor-specific T cell response. The with antitumor responses, including IL-2, IFN-γ, perforin, TNF-α,
paucity of shared tumor-specific antigens is in contrast to other and Granzyme B (Fig. 1C).
cancers, such as melanoma, where some patients mount sponta- A series of activation-associated, cell surface markers have
neous responses against the melanocyte differentiation antigen, recently been described to identify, characterize and utilize nat-
MART-1, which can be used to rapidly identify tumor-specific T urally occurring tumor-specific T cells in human tumors. CD137,
cells in melanoma patients using peptide/MHC detection agents PD-1, CD103, and CD39 are most commonly utilized as biomark-
[9]. Furthermore, many cancers including ovarian cancer, have ers of TILs with tumor-specificity [10–13, 18]. MC5, which highly
limited numbers of T cells that naturally respond to tumor- expresses effector molecules, was enriched for TILs expressing
specific antigens, making their examination challenging. Identi- high levels of CD137 as well as the co-stimulatory receptor
fying robust biomarkers for tumor-specific TILs can address this OX40, another TNFR family member upregulated upon T cell
issue. activation. MC5 moderately expressed CD103, PD-1, and CD39,
Various biomarkers are used to detect endogenous tumor- as well as activation markers CD69 and CD25 (Fig. 1D), indi-
specific T cells from solid tumors, such as the co-stimulatory cating that cells in MC5 are enriched for an activated T cell
receptor CD137 (also known as 4-1BB and TNFRSF9), the population.
negative immunoregulatory receptor PD-1, the lymphocyte-
retention mediating integrin CD103, and the co-expression of
both the ectonucleotidase CD39 and CD103 [10–13]. Iden-
tifying a singular, accurate biomarker for tumor-specific TILs CD137+ TILs preferentially express effector molecules
would streamline downstream research and clinical applications, and co-express biomarkers of tumor-specificity
but it is unknown which singular biomarker is most effec-
tive at identifying the tumor-specific TIL subset, as a direct To gain a further understanding of which biomarkers are most
comparison of these reported biomarkers has not been per- selective in identifying TILs expressing effector molecules within
formed. Addressing this knowledge gap is particularly impor- human cancer, we examined viSNE plots of activation and
tant, because TILs frequently co-express these markers and each tumor-specific biomarkers, which revealed the heterogeneity of
biomarker can be differentially expressed across the TIL pop- their expression patterns. Similar to what was observed in the
ulation, therefore, a biomarker comparison is needed to iden- metaPhenoGraph heat map results (Fig. 1D), CD137 expression
tify the marker that most accurately discerns tumor-specific TILs was primarily detected in the MC5 region, was expressed by both
[14, 15]. CD4+ and CD8+ TILs, and had co-expression of OX40, CD103,
Here, we compared the expression of CD137, PD-1, CD103, CD39, and PD-1 (Fig. 2A). PD-1 and CD69 expression were
and CD39 on TILs in human ovarian cancer, as these are leading common, broadly distributed, with overlapping expression of
biomarkers used to identify tumor-specific TILs. We hypothesized CD25, OX40, CD103, CD39, and CD137. CD25 and OX40 expres-
that a comparative interrogation of TILs in human tumors would sion were dominated by CD4+ TILs and commonly co-expressed
reveal which biomarker is most discriminating for tumor-specific with CD39, while CD103+ TILs were mainly CD8+ , a portion
TILs with autologous antitumor activity. of which expressed CD39. Overall, few TILs expressed effector

© 2021 Wiley-VCH GmbH www.eji-journal.eu

25
Eur. J. Immunol. 2022. 52: 96–108

Figure 1. A small population of patient TILs positive for activation, tumor-specific markers also co-express effector molecules. CyTOF was per-
formed on human ovarian tumor digests and analyzed using metaPhenoGraph. Experiment was repeated twice for a total of 5 samples (A) Metaclus-
tering analysis identified 7 metaPhenoGraph metaclusters for ovarian cancer patient CD3+ CD45+ TILs (n = 5). (B) Bar plot represents metacluster
frequency per patient. Error bars represent the 95% confidence interval (n = 5). (C) metaPhenoGraph cluster results are represented as a heatmap,
showing expression of Ki67, CD4, CD8, and effector markers (n = 5). (D) metaPhenoGraph heatmap displaying activation/tumor-specific markers
(n = 5).

molecules, such as IFN-γ, IL-2, and TNF-α, and their MMI was low, CD103+ CD39+ TILs to CD137+ TILs, CD137 expression was
compared to the level of activation and tumor-specific biomark- more selective for identifying total CD3+ and CD8+ TILs express-
ers. However, the few TILs that expressed effector molecules such ing effector molecules (Supporting Information Fig. 2A,B) with
as IFN-γ, IL-2, and TNF-α, were positive for CD137 in the MC5 no differences observed when comparing CD4+ TILs (data not
region, suggestive of CD137+ TIL polyfunctionality, and CD137 shown).
expression was more focal than other tumor-specific biomarkers Although CD137+ TILs exhibited the highest expression of
(Fig. 2A). Since viSNE plot analyses indicated that CD137 expres- effector molecules, the frequency of these cells was low (mean =
sion overlapped more with effector molecule expression than 4.1%, 95% CI = 1.87 to 6.36) compared to TILs express-
other biomarkers, we compared effector molecule expression ing other biomarkers (Fig. 3A). Since viSNE and PhenoGraph
within the CD137+ TIL population to expression in TILs express- analyses (Fig. 1) revealed that CD137+ TILs often co-express
ing other tumor-specific and activation markers (Supporting tumor-specific biomarkers, we next examined the frequency of
Information Fig. 1). Generally, CD137+ TILs exhibited the greatest CD137+ TILs within TIL populations expressing other tumor-
frequency of cells expressing IFN-γ, TNF-α, Granzyme B, perforin, specific biomarkers using biaxial gating (Fig. 3B). CD137+ TILs
and IL-2, compared to other biomarkers expressing TILs (Fig. 2B). commonly co-expressed PD-1 (mean = 54.9%, 95% CI = 39.47
CD137+ TILs had greater expression of IFN-γ (P-value = 0.0008), to 70.31]), CD103 (mean = 37.6%, 95% CI = 24.64 to 48.78),
Granzyme B (P-value = 0.01), perforin (P-value = 0.003), and and CD39 (mean = 76.8%, 95% CI = 64.55 to 88.95). In
IL-2 (P-value = 0.002) than CD103+ TILs, but similar levels of contrast, only a small portion of PD-1+ (mean = 6.2%, 95%
TNF-α expression (Fig. 2B). CD137+ TILs and OX40+ TILs were CI = 4.36 to 8.07), CD103+ (mean = 6.2%, 95% CI = 3.13
similar with the exception of CD137+ TILs expressing greater fre- to 9.17), or CD39+ (mean = 6.7%, 95% CI = 3.45 to 10.02)
quencies of IFN-γ (P-value = 0.008) and Granzyme B (P-value = TILs co-expressed CD137. These results, combined with effec-
0.01; Fig. 2B). While this study focuses on comparing single tor molecule expression data (Fig. 2), indicate that CD137 is
biomarkers, dual expression of CD103+ CD39+ was reported the more selective marker for identifying tumor-specific TILs
to identify CD8+ tumor-specific TILs [13]. When comparing (Fig. 3C).

© 2021 Wiley-VCH GmbH www.eji-journal.eu

26
Eur. J. Immunol. 2022. 52: 96–108 Systems immunology

Figure 2. TILs positive for CD137 have enhanced expression of effector molecules compared to other markers indicative of tumor-specificity.
CyTOF analysis was performed on human ovarian tumor digests and analyzed via viSNE (A) or traditional biaxial gating (B). (A) viSNE plots of
tumor-specific markers used in the literature (CD137, CD103, PD-1, CD39), activation markers (CD25, CD69), a TNFR family member (OX40), and
effector molecules IFN-γ and TNF-α. CD4 and CD8 TILs are represented as well. Experiment was conducted twice for a total of 5 samples. (B) Plots
comparing frequency of effector molecule expression between activation and markers of tumor-specificity (n = 15). Experiment was independently
performed four times, with the exception of CD39 where the experiment was repeated three times. The Student’s two-tailed, paired t-test was run
to determine statistical significance. NS represents a P-value >0.050, (*) represents a P-value ≤ 0.050, (**) represents a P-value <0.01, (***) represents
a P-value < 0.001, (****) represents a P-value < 0.0001, boxplots depict the median, quartiles, with whiskers representing the min and max, error
bars are 95% confidence interval.

CD137+ TILs are a subset of PD-1+ , CD103+ , and (P-value = 0.001), CD103+ (P-value < 0.001), and PD-1+ (P-
CD39+ TILs that exhibit antitumor activity value = 0.002) TILs when CD137+ TILs were selectively gated
out (Supporting Information Fig. 2C) prior to analysis in Fig. 4A.
We next investigated whether the CD137+ TIL subset contained This effect was also observed in TILs expressing CD25 (P-value
within other biomarker populations are enriched for effector = 0.001), CD69 (P-value = 0.001), or OX40 (P-value = 0.003;
molecules. Decreased IFN-γ expression was observed in CD39+ Fig. 4A). Granzyme B expression similarly decreased (Fig. 4B),

© 2021 Wiley-VCH GmbH www.eji-journal.eu

27
Eur. J. Immunol. 2022. 52: 96–108

Figure 3. CD137+ TILs are rare and highly co-express PD-1, CD103, and CD39. Human ovarian tumor digests were interrogated using CyTOF and
analyzed using traditional biaxial gating. Experiment was conducted a total of four times, with the exception of CD39 which was performed three
times independently. (A) Frequency of activation and tumor-specific markers within CD3+ CD45+ TILs (n = 15). (B) Co-expression patterns of CD137+ ,
PD-1+ , CD103+ , and CD39+ TILs (n = 15 with the exception of CD39+ TIL co-expression for PD-1, n = 10). (C) Schematic of CD137, PD-1, CD103,
and CD39 predicted T cell expression according to phenotypic findings. The Student’s two-tailed, paired t-test was run to determine statistical
significance. NS represents a P-value > 0.050, (*) represents a P-value ≤ 0.050, (**) represents a P-value < 0.01, (***) represents a P-value < 0.001, (****)
represents a P-value < 0.0001, error bars are 95% CI with center values representing the mean.

leading us to hypothesize that the CD137+ TIL subset may gous tumor cells exposure in three independent donor samples.
account for the reactivity observed in other biomarker-expressing Production of IFN-γ by CD137+ TILs upon autologous tumor
tumor-specific TIL populations [10–13]. co-culture was HLA-dependent, indicating tumor-antigen speci-
We next tested whether functional reactivity of TILs was ficity, as IFN-γ decreased upon HLA blocking of MHC class I
restricted to the CD137+ TIL subset in co-culture assays where and class II with antibodies (Fig. 4C). In all tested TIL sam-
CD137+ TILs were first sorted out of the bulk TIL, and then ples, the CD137+ subset secreted IFNγ levels twice as high as
other biomarker expressing TIL subsets were sorted prior to co- that of unstimulated TILs alone. These results indicate that effec-
culture with autologous tumor cells ( Supporting Information tor molecule expression is enriched within the CD137+ TIL frac-
Fig. 2D). Compared to the PD-1+ , CD103+ , and CD39+ TIL tion, and that CD137+ TILs account for the majority of antitu-
populations depleted of CD137+ cells, the CD137+ TIL sub- mor reactivity observed within PD-1+ , CD39+ , and CD103+ TIL
set produced the highest levels of IFN-γ in response to autolo- populations.

© 2021 Wiley-VCH GmbH www.eji-journal.eu

28
Eur. J. Immunol. 2022. 52: 96–108 Systems immunology

Figure 4. Removal of CD137+ T cells decreases effector molecule production in other biomarker subsets. (A and B) CyTOF analysis was run on
human ovarian tumor digests and analyzed via biaxial gating. (C) TIL subsets were co-cultured with autologous tumor cells and supernatants were
analyzed after 24 h by LEGENDplax. (A) Expression of IFN-γ and (B) Granzyme B within CD137+ and CD137− subpopulation within tumor-specific
and activation markers (n = 15). Experiment was conducted four times, for a total of total of 15 samples, with the exception of the CD39 plot
where the experiment was repeated three times for a total of 15 samples. Representative gating is shown in Supporting Information Figure 2C.
(C) IFN-γ secretion following autologous tumor co-culture in three different ovarian patient samples. Blue dashed and blue number on the y-axis
indicates the lowest sensitivity of the assay according to the standard curve. Assay flow setup is represented in Supporting Information Figure 2D.
Experimented was conducted three times, with one patient ran at a time. The Student’s two-tailed, paired t-test was run to determine statistical
significance. NS represents a P-value > 0.050, (*) represents a P-value ≤ 0.050, (**) represents a P-value < 0.01, (***) represents a P-value < 0.001, (****)
represents a P-value < 0.0001, error bars are 95% confidence interval with center values representing the mean.

© 2021 Wiley-VCH GmbH www.eji-journal.eu

29
Eur. J. Immunol. 2022. 52: 96–108

in CD4+ TILs. CD8+ TILs contained greater frequencies of cells

expressing Granzyme B (P-value = 0.03), and no difference was
detected in perforin expression (Fig. 5C). These results support
the notion that both CD4+ and CD8+ TILs can express effector
molecules, which can be divergent and together may play integral
roles in immune responses against tumor cells.

Tumor-specific TILs display a phenotype indicative of

restorable exhaustion

Since our findings indicate that CD137+ TILs express effec-

tor molecules and other molecules indicative of activation, we
queried whether these tumor-specific TILs displayed features of
exhaustion, which is commonly associated with chronic tumor-
antigen stimulation [19]. We examined to what degree TILs in
meta-cluster 5 (MC5) (Figs. 1 and 2), which harbored the high-
est frequency of CD137+ and effector molecule-expressing TILs,
exhibit phenotypic hallmarks of exhaustion. T cells in the MC5
population expressed multiple markers indicative of exhaustion,
specifically CTLA-4, Tim-3, PD-1, CD39, CD244, EOMES, Lag-3,
TIGIT, and CD160. (Fig. 6A). MC5 TILs expressed PD-1, albeit at
Figure 5. Both CD4+ and CD8+ TILs express markers indicative of anti- overall lower levels than MC6. MC5 TILs uniquely co-expressed
tumor reactivity. Human ovarian tumor digests were interrogated by PD-1 and the costimulatory molecule CD28, whose signaling is
CyTOF and analyzed via biaxial gating. Experiment was independently
run four times, with the exception for CD39 which was repeated three
required for rescue of CD8+ T cell activity in anti-PD-1 ther-
times. (A) Frequency of CD4+ and CD8+ TILs in live CD3+ CD45+ TILs apy for cancer [20]. Exhaustion-associated marker expression in
(n = 19). (B) Frequency of tumor-specific markers within CD4+ and CD8+ CD137+ TILs was compared to other TIL populations by examin-
TILs. Sample size per group: CD137 (n = 19), CD39 (n = 14), CD103, PD-
1, CD25, CD69, and OX40 (n = 18). (C) Frequency of effector molecules
ing TIGIT, EOMES, and CD39 expression in CD137+ or CD137−
within CD4+ and CD8+ TILs. The Student’s two-tailed, paired t-test subsets. CD137+ TILs expressed higher levels of the exhaustion-
was run to determine statistical significance. NS represents a P-value associated markers TIGIT (P-value < 0.001), EOMES (P-value
> 0.050, (*) represents a P-value ≤ 0.050, (**) represents a P-value < 0.01,
(***) represents a P-value < 0.001, (****) represents a P-value < 0.0001,
< 0.001), and CD39 (P-value < 0.001), compared to CD137− TILs
error bars are 95% confidence interval with center values representing (Fig. 6B). Since the aforementioned markers can be upregulated
the mean. by both activated and exhausted T cells, we assessed whether
CD137+ TILs are skewed towards a EOMEShi T-betdim phenotype
Both CD4+ and CD8+ TILs express markers of associated with dampened effector functions [21] or toward a
antitumor reactivity more functional EOMESdim T-bethi phenotype. CD137+ TILs were
more skewed towards an EOMEShi T-betdim (P-value = 0.004)
Having analyzed co-expression and effector profiles of phenotype than their CD137− counterparts, supporting the notion
tumor-specific marker expressing populations within overall that CD137+ TILs are exhausted (Fig. 6C, D). As CD137+ TILs
CD3+ CD45+ TILs, we next examined the biomarker profiles of appear exhausted but also harbor tumor-specific TILs that express
CD4+ or CD8+ TIL subsets in ovarian cancer samples. There were effector molecules and co-express CD28, our results suggest
more CD4+ TILs (mean = 49.6%, 95% CI = 43.43 to 55.86) than that CD137+ TILs have the greatest potential for reinvigoration
CD8+ TILs (P-value = 0.02, mean = 34.9%, 95% CI = 27.88 to [20]; however, studies designed to disentangle functional T cell
41.85) (Fig. 5A). CD137 expression within the CD4+ (mean = exhaustion from activity in CD137-expressing TILs would be nec-
5.9%, 95% CI = 2.12 to 9.75) and CD8+ (mean = 3.32%, 95% CI essary to validate this supposition.
= 5.46 to 1.19) TIL subsets did not statistically differ, suggesting
that both CD4+ and CD8+ T cells harbor tumor-specific TILs and
that both subsets may contribute to antitumor activity in ovarian Discussion
cancer. More CD8+ TILs expressed CD103 (P-value < 0.001)
and CD69 (P-value = 0.01). A higher percentage of CD4+ TILs TILs are a heterogeneous population of immune cells that can
expressed CD25 (P-value < 0.001) and OX40 (P-value = 0.01), differ in specificity, differentiation, and function. Biomarkers
but there was no significant differences in CD39 or PD-1 expres- that identify endogenous tumor-specific TIL subsets are fun-
sion (Fig. 5B). When comparing effector molecule expression, an damental to immunobiology research, studying mechanisms of
increased frequency of IFN-γ (P-value = 0.04), TNF-α (P-value endogenous antitumor immunity, isolating tumor-specific T-cell
= 0.01), and IL-2 (P-value = 0.01) expressing cells was detected receptors, and optimizing cellular therapies [5–8]. We observed

© 2021 Wiley-VCH GmbH www.eji-journal.eu

30
Eur. J. Immunol. 2022. 52: 96–108 Systems immunology

Figure 6. CD137+ TILs within the ovarian cancer tumor microenvironment have a phenotype indicative of exhaustion. CyTOF analysis was per-
formed on human ovarian tumor digests and analyzed via biaxial gating. (A) metaPhenoGraph heatmap (same samples and setup as in Figure 1)
displaying markers mainly of activation, co-stimulation, and exhaustion. Experiment was conducted twice (n = 5). (B) Frequencies of exhaus-
tion markers TIGIT, EOMES, and CD39 in CD137+ versus CD137− TILs (n = 11). Experiment was repeated three times. (C) Representative gating of
EOMEShi T–betlo and EOMEShi T-betlo in CD137+ and CD137− live CD3+ CD45+ TILs. (D) EOMEShi T–betlo and EOMEShi T-betlo frequencies in CD137+
and CD137− TILs (n = 7). Experiment was performed independently twice. The Student’s two-tailed, paired t-test was run to determine statistical
significance. NS represents a P-value > 0.050, (*) represents a P-value ≤ 0.050, (**) represents a P-value <0.01, (***) represents a P-value < 0.001, (****)
represents a P-value < 0.0001, error bars are 95% confidence interval with center values representing the mean.

that TILs expressing effector molecules often co-expressed other previous findings from our laboratory [9, 10], and later studies
biomarkers used to identify tumor-specific TILs. Earlier studies of that used CD137 to enrich tumor-specific TILs [7,10,22]. Our find-
TILs expressing a single biomarker reported levels of secondary ings contradict results reported by Gros and colleagues showing
biomarker co-expression, but a direct comparison between var- that both PD-1+ and CD137+ TIL subsets were tumor-reactive
ious biomarker-expressing TIL subsets had yet to be conducted but with PD-1 better identifying tumor-reactive T cells [11].
[10–13]. We found that a small subset of PD-1+ , CD103+ , and Interesting, activation-induced expression of CD137 was used to
CD39+ TILs reproducibly co-express CD137. In contrast, most define tumor-reactivity in many of the assays used in that study.
CD137+ TILs highly co-express the aforementioned biomark- The discrepancy between our findings and those reported by Gros
ers, and preferentially express effector molecules, indicating that et al. may be explained by differences in the cancer type stud-
CD137 more selectively identifies tumor-specific TILs. Further, ied as well as the methodology applied. Gros et al. solely focused
removing CD137+ TILs from other biomarker-expressing TIL sub- on CD8+ TILs and did not include CD4+ TILs. In contrast, the
sets reduced their functional activity in response to autologous present study, and our previous study that first defined CD137 as a
tumor stimulation, indicating that while PD-1, CD103, and CD39 biomarker for tumor-specific TILs [10] included CD4+ TILs in the
markers can be used to identify tumor-specific TILs, CD137 analysis. This alone does not account for the discrepancy, since
expression is a more discriminatory tumor-specific TIL biomarker. CD137 still served as a better biomarker for tumor-specific CD8+
The finding that CD137 expression is a highly selective marker TILs. Identifying endogenous tumor-antigen-specific TIL biomark-
for endogenous tumor-specific TIL identification is supported by ers in patients has been heavily CD8+ T-cell-centric [11–13, 23],

© 2021 Wiley-VCH GmbH www.eji-journal.eu

31
Eur. J. Immunol. 2022. 52: 96–108

but there is growing appreciation for the role of CD4+ T cells in CD137/OX40 bispecific antibody [35] may be promising to test in
promoting antitumor immunity and immunotherapy efficacy [24– an ovarian cancer model, as we observed that effector molecule
27]. This is emphasized by our findings that CD4+ TILs dominate expressing CD137+ TILs also co-expressed OX40 (Fig. 2A). Future
the ovarian tumor environment and have equivalent expression studies are needed to determine if CD137+ TILs are reinvigo-
of CD137 as CD8+ TILs. Also, with the exception of Granzyme rated by anti-PD-1 therapy, whether they require CD28 signal-
B, CD4+ TILs had either equivalent or greater positivity for IFN- ing, and how they contribute to successful immune checkpoint
γ, TNF-α, perforin, and IL-2. Our results support the idea that blockade monotherapy or combinatorial immunotherapy strate-
both CD8+ and CD4+ TILs have integral roles in driving antitu- gies [35, 36].
mor immune responses and may have divergent antigen-specific Collectively, this work clarifies the differential expression of
responses. biomarkers for tumor-specific TILs and demonstrates that CD137
A separate study by Duhen et al. demonstrated that co- is a more selective biomarker for identifying naturally occurring,
expression of CD39 and CD103 TILs can identify tumor-specific tumor-specific TILs than PD-1, CD103, or CD39 within human
TILs within solid tumors. Similar to Gros et al., the work ovarian tumors. This work corroborates our original finding that
focused on CD8+ TILs [13]. Supporting our finding that CD137+ CD137 accurately identifies tumor-specific TILs10 . Furthermore,
TILs often co-express other commonly used tumor-specific TIL our findings explain why the addition of an agonistic antibody
biomarkers, both Gros et al. and Duhen et al., used CD137 upreg- to TIL cultures results in preferential expansion of tumor-specific
ulation as a measure to assess tumor-cell recognition by PD-1+ TILs in melanoma [37]. We acknowledge there are limitations to
or CD39+ CD103+ CD8+ TILs in co-culture experiments. Notably, this analysis. Our study entirely used ovarian cancer specimens,
only a subset of enriched PD-1+ or CD39+ CD103+ CD8+ TILs and results may differ in other cancer types. Also, due to limited
upregulated CD137 expression after autologous tumor recogni- cell numbers, we were unable to independently test CD4+ and
tion. Unlike Gros et al. and Duhen et al. studies, we examined CD8+ TILs for TIL subset reactivity, or test restorable exhaustion
TILs from tumor digests without the addition of cytokines, estab- on PD-1+ T cells. Furthermore, PD-1 blockade has low efficacy in
lishment of T cell clones, or bulk-expansion. It bears consideration in vitro assays, and would require sophisticated in vivo models
that this methodology may better preserve TIL natural reactivities and large cell numbers. Nevertheless, we conclude that this work
to autologous tumor antigens with minimal manipulation of TIL disentangles the differential expression of tumor-specific biomark-
biomarker expression. ers by TILs and demonstrates that CD137 is an ideal singular
Immune checkpoint blockade has shown great promise in biomarker for identifying tumor-specific TILs, which provides a
numerous solid tumors, and successful antitumor responses are deeper understanding of human TILs that may pave a route
thought to rely upon reinvigorated responses by tumor- specific towards improving immunotherapeutic strategies for cancer.
T cells [19, 28, 29]. The phenotypic profile of CD137+ TILs sug-
gests that they have potential for reinvigoration via checkpoint
blockade. CD137+ TILs highly expressed multiple co-inhibitory
receptors, including PD-1, and were skewed towards a pheno- Materials and methods
type characteristic of exhausted T cells [21] and co-expressed
CD28. Expression of CD28 by CD137+ TILs is important because Tumor Samples
restoring exhausted T cell function is dependent on CD28 co-
stimulation [20, 30]. However, many cancers, including ovarian Viably frozen, human high-grade serous ovarian tumor samples
cancer, have low response rates to PD-1/PDL1 blockade [31]. were purchased from the Penn Ovarian Cancer Research Center
Our data may suggest that one potential explanation is that (OCRC) Tumor BioTrust Collection. Ethics statement: All donor
most patients have too few CD137+ TILs to reinvigorate for samples used in this study were de-identified and approved for
an effective antitumor response. It is intriguing to hypothesize use by the UPenn Institutional Review Board (IRB 702679, UPCC
that the response rate to PD-1/PDL1 blockade may be increased 17909). Sex and weight are not a biological variable as all
by promoting CD28 signaling to TILs, such as through CTLA-4 tumor samples are from females. As samples are de-identified,
blockade. Both CTLA-4 and CD28 bind to CD80 and CD86 on age and weight are not known. Surgically resected tumors were
antigen-presenting cells, but CTLA-4 binds CD80 and CD86 with procured from the operating room in an aseptic manner. Tis-
greater affinity and avidity than CD28, enabling it to outcom- sue was mechanically processed into fragments and added to
pete CD28 for these ligands. The response rate to anti-PD-1 anti- an enzyme digest solution. A 10× stock solution of the enzyme
body treatment in ovarian cancer nearly triples when a CTLA- digest buffer contains 2 mg/mL collagenase (Sigma–Aldrich) and
4 blocking antibody is added to the treatment regimen [32]. 0.3 kU/mL DNase I Type IV (Sigma–Aldrich); solution was diluted
Furthermore, agonizing CD137 may aid in promoting antitumor to a 1× solution with RPMI 1640 at the time of digestion. Tis-
responses in patients, and although CD137 agonism in the clinic sue was incubated in the enzyme digest buffer overnight at room
has had toxicities [33, 34], dual bispecific antibodies that agonize temperature on a rotator. Dissociated tumor tissue was subse-
CD137 are being developed in order to enhance T cell prolifer- quently filtered through sterile 100μm nylon mesh, centrifuged,
ation and antitumor activity in human cancer without the safety and washed twice with dPBS (Dulbecco’s Phosphate Buffered
limitations observed in the clinic [35, 36]. The recently developed Saline). Resultant tumor cell digests were cryopreserved in 10%

© 2021 Wiley-VCH GmbH www.eji-journal.eu

32
Eur. J. Immunol. 2022. 52: 96–108 Systems immunology

dimethyl sulfoxide (DMSO) (Sigma Aldrich) and human serum were subsequently stained with Zombie aqua (BioLegend Cat#
(Valley Biomedical, Inc., Product #HS1017). Samples were frozen 423102) for 10 minutes to discriminate live and dead cells. Sam-
at -80°C and banked at -150°C until further use. ples were washed twice to remove Zombie aqua, then incubated
at 4°C for 30 min in 50 μl of an antibody cocktail to label human
surface markers. Following surface staining, samples were washed
Mass Cytometry staining three times. Samples were sent to the Flow Cytometry Facility at
the Wistar Institute for fluorescent-activated cell sorting (FACS)
CyTOF antibodies were purchased from Fluidigm as pre- on a MoFlo Astrios or to the Flow Cytometry Core at the Chil-
conjugated metal tagged antibodies or were conjugated in-house dren’s Hospital of Philadelphia and sorted on an Aria, and adhered
using the Maxpar Fluidigm kit and protocol. All antibodies were to the guidelines for the use of flow cytometry and cell sorting in
titrated to determine optimal concentrations for staining samples. immunological studies [39]. All antibodies were purchased from
The panel used to initially investigate tumor-specific markers, BioLegend. For all analyses, singlets were detected using FSC-
before inclusion of CD39 in the aforementioned panel, had the H versus FSC-A followed by SSC-H versus SSC-A. Cells negative
following surface markers: CD3, CD45, CD4, CD8, CD244, CD69, for Zombie aqua, were identified as live cells. Anti-human-anti-
OX40, Lag-3, CD103, Tim-3, TIGIT, PD-1, CD137, CD28, CD127, CD3-PerCpCy5.5 (BioLegend Cat# 317336, RRID:AB_2561628)
CD27, GITR, CD25, HLA-DR, and CD160. Intracellular antibodies was used to detect T cells and the following anti-human antibod-
included: CTLA-4, pStat5, IL-17A, IL-2, IFNg, Granzyme B, Ki67, ies were used to identify T cell subsets CD137+ , PD-1+ CD137- ,
and Perforin. No additional polyclonal stimulation or protein CD39+ CD137- , CD103+ CD137- : anti-CD137-PeCY7 (BioLegend
transport inhibitors were added to preserve the natural pheno- Cat# 309818, RRID:AB_2207741), anti-CD103-BV605 (BioLe-
type and activation state of the T cells. We subsequently designed gend Cat# 350218, RRID:AB_2564283), anti-CD39-APC (BioLe-
a panel to include CD39 and all other tumor-specific markers of gend Cat# 328210, RRID:AB_1953234), and anti-PD-1-APCCy7
interest. The following panel included CD39 and was used for (BioLegend Cat# 329922, RRID:AB_10933429).
downstream viSNE, metaPhenoGraph, and biaxial analysis. Anti-
human surface markers for the panel were: CD3, CD45, CD4,
CD8, CD103, PD-1, OX40, CD39, CD69, CD25, CD137, CD27, Mass cytometry biaxial analyses
Tim-3, CD127, CD28, CD244, CD5, Lag-3, TIGIT, HLA-DR, and
CD160. Intracellular markers included: Ki67, IL-17A, IL-2, IFN-γ, Traditional biaxial analysis, on bead-normalized fcs files, was per-
IL-6, Perforin, pStat5, TNF-α, Granzyme B, CTLA-4, and EOMES. formed using Flowjo V10 software (FlowJo, RRID:SCR_008520).
The initial panel to compare CD39 and CD137 positive TILs had Intact single cells were identified using event-length and Iridium.
the following surface antibodies interrogated: CD3, CD45, CD4, Cells were live-gated according to 127IdU and mm-DOTA, where
CD8, CD137, CD39, CD25, HLA-DR, and CD127. Intracellular dead cells are positive for mm-DOTA. CD3 and CD45 positivity
antibodies detected were: IL-2, pStat5, EOMES, T-bet, IL-17A, identified T-cells. Sequential gating analysis was performed for
IFN-γ, Granzyme B, Ki67, and Perforin. The last panel used in this all analyzed markers. The resulting values were used to determine
study was designed to focus on TIL exhaustion. Surface antibodies population frequencies.
used were: CD3, CD45, CD4, CD8, OX40, CD103, TIGIT, CD137,
CD39, CD25, CD3, HLA-DR, and CD127. Intracellular antibod-
ies were: IL-2, pStat5, EOMES, T-bet, and Ki67. For all panels, viSNE and metaPhenoGraph analyses
cell identifier stain Iridium191/193, live identifier 127IdU (Flu-
idigm) were used. To discriminate dead cells, cisplatin purchased High-dimensional analysis was conducted using the algorithm
from Fluidigm or dead stain maleimido-mono-amine-DOTA (mm- viSNE, which uses the Barnes-Hut t-SNE (bh-SNE) implementa-
DOTA) from Macrocyclics was used. Viably frozen ovarian human tion, from cyt a visualization tool written in Matlab (R2016b,
tumor digests obtained from the Tumor BioTrust collection, pro- MATLAB,RRID:SCR_001622) downloaded in 2015 and avail-
cessed as described in the Tumor Samples section above, were able at https://www.c2b2.columbia.edu/danapeerlab/html/cyt-
thawed in batches and stained for CyTOF following the same download.html. Live, single, CD3+ CD45+ CD137+/- exported fcs
methodology as Bengsch et al. [38]. Data acquisition was per- data from five donor samples were imported into cyt, arcsinh5-
formed on a CyTOF Helios (Fluidigm CyTOF Helios Mass Cytome- transformed, and run as described by Amir et al., 2013 [16] to
ter, RRID:SCR_019916) by the CyTOF Mass Cytometer Core at create viSNE plots. The following parameters were used for bh-
UPenn. The core performed bead-based normalization for all SNE mapping analysis: Ki67, IL-17A, IL-2, IFN-γ, CD103, PD-
samples. 1, IL-6, OX40, CD39, Perforin, CD69, CD4, CD8, pStat5, TNF-α,
GITR, CD25, Granzyme B, and CD137. The PhenoGraph algo-
rithm was run, as described by Levin et al. [17], with the near-
Fluorescent-activated cell sorting est neighbor input of k = 30 and a Euclidean distance metric.
Markers used for PhenoGraph clustering were the following: Ki67,
Tumor samples were thawed and washed twice with staining IL-17A, IL-2, IFN-γ, CD103, PD-1, IL-6, OX40, CD39, Perforin,
buffer (PBS, 5% fetal bovine serum) to remove DMSO. Samples CD69, CD4, CD8, pStat5, TNF-α, GITR, CD25, Granzyme B, and

© 2021 Wiley-VCH GmbH www.eji-journal.eu

33
Eur. J. Immunol. 2022. 52: 96–108

CD137. PhenoGraph was metaclustered, as described by Levine their aid and expertise in sorting cells. This work was supported in
et al., using a k = 15 and a Euclidean distance metric. viSNE, part by an NIH/NCI pilot grant P50 CA228991 awarded to Daniel
PhenoGraph, metaPhenoGraph plots, and heatmaps were created J. Powell Jr., and partly by BMS grant CA186-113 awarded to Jes-
by cyt. sica A. Chacon and Daniel J. Powell Jr. The funding sources had
no role in the design, collection, analysis, interpretation, writing,
or decision to submit the manuscript for publication.
Co-culture experiment

The following live T cell subsets were FACS sorted from thawed Data availability statement: The data supporting the findings
cryopreserved patient tumor samples: CD137+ , CD39+ CD137− , of this study are available from the corresponding author upon
CD103+ CD137− , and PD-1+ CD137- using the staining proto- reasonable request.
col specified in the FACs sorting section. T cells were rested
overnight in RPMI 1640 media supplemented with 10% FBS Peer review: The peer review history for this article is available
and supernatants were collected the following day. CD45+ cells at https://publons.com/publon/10.1002/eji.202149329
were depleted from the same patient sample to obtain CD45−
cells for co-culture using the EasySep Human CD45 Depletion
Kit from StemCell Technologies Cat# 17898. T cell subsets Ethics approval statement
were co-cultured, with 10ug/ml HLA-blocking Class I (BioLegend
Cat# 311402, RRID:AB_314871) & II (BioLegend Cat# 361702, De-identified human ovarian cancer samples were obtained from
RRID:AB_2563139) or isotype (BioLegend Cat# 400202) anti- the Penn Ovarian Cancer Research Center (OCRC) Tumor BioTrust
bodies, at a 1:2 ratio of T cells to autologous tumor cells in Collection (UPCC 17909, IRB 702679).
100ul of media in a 96-ubottom plate. The number of T cells
added to co-culture were as follows; 30,000 cells for sample 1743;
14,500 cells for 2304 and 6125 cells for 2399. Following 24 h co- Author contributions
culture, samples were spun down at 1300 rpm, and supernatants
were collected and stored at -80°C until use. To analyze cytokines M.A.E. and D.J.P were associated with conceptualization,
within the supernatants, the manufacture’s protocol of the LEG- methodology, project administration, and writing; D.J.P. was asso-
ENDplex Human CD8/NK Panel kit (BioLegend Cat# 740267) ciated with supervision and resources. D.J.P and J.C. were associ-
was followed, and two technical replicates were analyzed per ated with funding acquisition. M.A.E. was associated with formal
sample. analysis, visualization, investigation, and validation; D.K.O. was
associated with resources and writing; J.C. reviewed and edited
the final manuscript.
Statistical analysis
Conflict of interest: D.J.P. holds a patent on CD137 enrichment
for efficient tumor infiltrating lymphocyte selection (U.S. Pat. No.
The Student’s two-tailed, paired t-test was run to determine statis-
10,233,425) and receives fees for advisory services from InsTIL
tical significance. NS represents a P-value > 0.050, “*” represents
Bio on TIL therapy. All other authors have no commercial or finan-
a P-value ≤ 0.050, “**” represents a P-value < 0.01, “***” rep-
cial conflicts of interests.
resents a P-value < 0.001, “****” represents a P-value < 0.0001,
error bars represent 95% confidence Interval.

References
1 Zhang, L., Conejo-Garcia, J. R., Katsaros, D., Gimotty, P. A., Massobrio, M.,
Regnani, G., Makrigiannakis, A. et al., Intratumoral T cells, recurrence,
and survival in epithelial ovarian cancer. N. Engl. J. Med. 2003; 348: 203–
Acknowledgements: The authors acknowledge the Ovarian Can- 213.

cer Translational Center for Excellence in the Abramson Can- 2 Geng, Y., Shao, Y., He, W., Hu, W., Xu, Y., Chen, J., Wu, C. et al., Prognostic
cer Center for support of tumor banking operations through role of tumor-infiltrating lymphocytes in lung cancer: a meta-analysis.
Cell. Physiol. Biochem. Int. J. Exp. Cell. Physiol. Biochem. Pharmacol. 2015; 37:
the Tumor BioTrust Collection. The authors thank Dr. Bertram
1560–1571.
Bengsch and the E. John Wherry laboratory at the University
3 Chen, M.-L., Pittet, M. J., Gorelik, L., Flavell, R. A., Weissleder, R., von
of Pennsylvania, for aid in the development of CyTOF antibody
Boehmer, H. and Khazaie, K., Regulatory T cells suppress tumor-specific
panels and insights on data analysis, as well Takuya Ohtani of CD8 T cell cytotoxicity through TGF-beta signals in vivo. Proc. Natl. Acad.
the UPenn CyTOF Core for running mass cytometry samples. The Sci. U. S. A. 2005; 102: 419–424.
authors also acknowledge the personnel of the Wistar Institute’s 4 Rabinovich, G. A., Gabrilovich, D. and Sotomayor, E. M., Immunosuppres-
Flow Cytometry Facility and the Flow Cytometry Core Laboratory sive strategies that are mediated by tumor cells. Annu. Rev. Immunol. 2007;
at the Children’s Hospital of Philadelphia Research Institute for 25: 267–296.

© 2021 Wiley-VCH GmbH www.eji-journal.eu

34
Eur. J. Immunol. 2022. 52: 96–108 Systems immunology

5 Rosenberg, S. A., Restifo, N. P., Yang, J. C., Morgan, R. A. and Dudley, M. E., 22 Seliktar-Ofir, S., Merhavi-Shoham, E., Itzhaki, O., Yunger, S., Markel,
Adoptive cell transfer: a clinical path to effective cancer immunotherapy. G., Schachter, J. and Besser, M. J., Selection of shared and neoantigen-
Nat. Rev. Cancer. 2008; 8: 299–308. reactive T cells for adoptive cell therapy based on CD137 separation. Front.
Immunol. 2017; 8: 1211.
6 Yossef, R., Tran, E., Deniger, D. C., Gros, A., Pasetto, A., Parkhurst, M.
R., Gartner, J. J. et al., Enhanced detection of neoantigen-reactive T 23 Wolfl M., Kuball J., Ho W. Y., Nguyen H., Manley T. J., Bleakley M., Green-
cells targeting unique and shared oncogenes for personalized cancer berg P. D. Activation-induced expression of CD137 permits detection, iso-
immunotherapy. JCI Insight. 3. lation, and expansion of the full repertoire of CD8+ T cells responding
to antigen without requiring knowledge of epitope specificities. Blood.
7 Parkhurst, M., Gros, A., Pasetto, A., Prickett, T., Crystal, J. S., Robbins, P. and
2007;110: (1):201–210.
Rosenberg, S. A., Isolation of T-cell receptors specifically reactive with
mutated tumor-associated antigens from tumor-infiltrating lymphocytes 24 Borst, J., Ahrends, T., Babała,
˛ N., Melief, C. J. M. and Kastenmüller, W.,
based on CD137 expression. Clin Cancer Res. 2017; 23: 2491–2505. CD4 + T cell help in cancer immunology and immunotherapy. Nat. Rev.
Immunol. 2018; 18: 635–647.
8 Leung, W. and Heslop, H. E., Adoptive immunotherapy with antigen-
specific T cells expressing a native TCR. Cancer Immunol. Res. 2019; 7: 528– 25 Friedman, K. M., Prieto, P. A., Devillier, L. E., Gross, C. A., Yang, J.
533. C., Wunderlich, J. R., Rosenberg, S. A. et al., Tumor-specific CD4+
melanoma tumor-infiltrating lymphocytes. J. Immunother. 2012; 35: 400–
9 Ramirez-Montagut, T., Andrews, D. M., Ihara, A., Pervaiz, S., Pandolfi, F.,
408.
Van Den Elsen, P. J., Waitkus, R. et al., Melanoma antigen recognition by
tumour-infiltrating T lymphocytes (TIL): effect of differential expression 26 Hunder, N. N., Wallen, H., Cao, J., Hendricks, D. W., Reilly, J. Z., Rodmyre,
of Melan-A/MART-1. Clin. Exp. Immunol. 2000; 119: 11–18. R., Jungbluth, A. et al., Treatment of metastatic melanoma with autol-
ogous CD4+ T cells against NY-ESO-1. N. Engl. J. Med. 2008; 358: 2698–
10 Ye, Q., Song, D.-G. G., Poussin, M., Yamamoto, T., Best, A., Li, C., Coukos,
2703.
G. et al., CD137 accurately identifies and enriches for naturally occurring
tumor-reactive T cells in tumor. Clin Cancer Res. 2014; 20: 44–55. 27 Tran E., Turcotte S., Gros A., Robbins P. F., Lu Y.-C., Dudley M. E., Wunder-
lich J. R. et al., Cancer Immunotherapy Based on Mutation-Specific CD4+
11 Gros, A., Robbins, P. F., Yao, X., Li, Y. F., Turcotte, S., Tran, E., Wun-
T Cells in a Patient with Epithelial Cancer. Science. 2014;344: (6184):641–
derlich, J. R. et al., PD-1 identifies the patient-specific CD8+ tumor-
645.
reactive repertoire infiltrating human tumors. J Clin Invest. 2014; 124: 2246–
2259. 28 Simon, S. and Labarriere, N., PD-1 expression on tumor-specific
T cells: Friend or foe for immunotherapy? Oncoimmunology. 2017; 7
12 Djenidi, F., Adam, J., Goubar, A., Durgeau, A., Meurice, G., de Mont-
(1): https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5739549/ [Accessed
préville, V. and Validire, P. et al., CD8+CD103+ tumor–infiltrating lympho-
April 7, 2020]. https://doi.org/10.1080/2162402X.2017.1364828.
cytes are tumor-specific tissue-resident memory T Cells and a Prognos-
tic factor for survival in lung cancer patients. J. Immunol. 2015; 194: 3475– 29 Thommen, D. S. and Schumacher, T. N., T Cell Dysfunction in Cancer.
3486. Cancer Cell. 2018; 33: 547–562.

13 Duhen, T., Duhen, R., Montler, R., Moses, J., Moudgil, T., Miranda, de 30 Kamphorst, A. O., Pillai, R. N., Yang, S., Nasti, T. H., Akondy, R. S., Wieland,
N. F., Goodall, C. P. et al., Co-expression of CD39 and CD103 identifies A., Sica, G. L. et al., Proliferation of PD-1+ CD8 T cells in peripheral blood
tumor-reactive CD8 T cells in human solid tumors. Nat. Commun. 2018; 9: after PD-1–targeted therapy in lung cancer patients. Proc. Natl. Acad. Sci.
2724. 2017; 114: 4993–4998.

14 Kamada, T., Togashi, Y., Tay, C., Ha, D., Sasaki, A., Nakamura, Y., Sato, 31 Longoria, T. C. and Eskander, R. N., Immune checkpoint inhibition: ther-
E. et al., PD-1+ regulatory T cells amplified by PD-1 blockade promote apeutic implications in epithelial ovarian cancer. Recent Patents Anticancer
hyperprogression of cancer. Proc. Natl. Acad. Sci. 2019; 116: 9999–10008. Drug Discov. 2015; 10: 133–144.

15 Lowther D. E., Goods B. A., Lucca L. E., Lerner B. A., Raddassi 32 Zamarin, D., Burger, R. A., Sill, M. W., Powell, D. J., Lankes, H. A.,
K., van Dijk D., Hernandez A. L. et al., PD-1 marks dysfunc- Feldman, M. D., Zivanovic, O. et al., Randomized Phase II trial of
tional regulatory T cells in malignant gliomas. JCI Insight. 2016;1: (5): nivolumab versus nivolumab and ipilimumab for recurrent or persis-
http://doi.org/10.1172/jci.insight.85935 tent ovarian cancer: an NRG oncology study. J. Clin. Oncol. 2020; 38: 1814–
1823.
16 Amir, ED l-AD, Davis, K. L., Tadmor, M. D., Simonds, E. F., Levine, J. H., Ben-
dall, S. C., Shenfeld, D. K. et al., viSNE enables visualization of high dimen- 33 Segal, N. H., Logan, T. F., Hodi, F. S., McDermott, D., Melero, I., Hamid, O.,
sional single-cell data and reveals phenotypic heterogeneity of leukemia. Schmidt, H. et al., Results from an Integrated Safety Analysis of Urelumab,
Nat Biotechnol. 2013; 31: 545–552. an Agonist Anti-CD137 Monoclonal Antibody. Clin. Cancer Res. Off. J. Am.
Assoc. Cancer Res. 2017; 23: 1929–1936.
17 Levine, J., Simonds, E., Bendall, S. and Davis, K., Data-driven phenotypic
dissection of AML reveals progenitor-like cells that correlate with prog- 34 Segal, N. H., He, A. R., Doi, T., Levy, R., Bhatia, S., Pishvaian, M. J., Cesari,
nosis. Cell 2015;162: 184–197. R. et al., Phase I study of single-agent Utomilumab (PF-05082566), a 4-
1BB/CD137 agonist, in patients with advanced cancer. Clin. Cancer Res. Off.
18 Lee, Y., Park, J., Park, S.-H. and Shin, E.-C., CD39+CD8+ T cells exhibit
J. Am. Assoc. Cancer Res. 2018; 24: 1816–1823.
a distinct phenotype among tumor-infiltrating tumor-antigenspecific
CD8+ T cells. J. Immunol. 2019; 202: 195.2–195.2. 35 Gaspar, M., Pravin, J., Rodrigues, L., Uhlenbroich, S., Everett, K. L., Woller-
ton, F., Morrow, M. et al., CD137/OX40 bispecific antibody induces potent
19 Pauken, K. E. and Wherry, E. J., Overcoming T cell exhaustion in infection
antitumor activity that is dependent on target coengagement. Cancer
and cancer. Trends Immunol 2015;36: 265–276.
Immunol. Res. 2020; 8: 781–793.
20 Kamphorst, A. O., Wieland, A., Nasti, T., Yang, S., Zhang, R., Barber, D. L.,
36 Lakins, M. A., Koers, A., Giambalvo, R., Munoz-Olaya, J., Hughes,
Konieczny, B. T. et al., Rescue of exhausted CD8 T cells by PD-1–targeted
R., Goodman, E., Marshall, S. et al., FS222, a CD137/PD-L1 tetrava-
therapies is CD28-dependent. Science 2017; 355: 1423–1427.
lent bispecific antibody, exhibits low toxicity and antitumor activ-
21 Wherry, E. and Kurachi, M., Molecular and cellular insights into T cell ity in colorectal cancer models. Clin. Cancer Res. 2020; 26: 4154–
exhaustion. Nat Rev Immunol. 2015; 15: 486–499. 4167.

© 2021 Wiley-VCH GmbH www.eji-journal.eu

35
Eur. J. Immunol. 2022. 52: 96–108

37 Chacon, J. A., Wu, R. C., Sukhumalchandra, P., Molldrem, J. J., Sar- Abbreviations:CI: confidence interval · MC: metacluster · TILs:
naik, A., Pilon-Thomas, S., Weber, J. et al., Co-stimulation through 4- tumor-infiltrating lymphocytes
1BB/CD137 improves the expansion and function of CD8(+) melanoma
tumor-infiltrating lymphocytes for adoptive T-cell therapy. PloS One. 2013;
Full correspondence: Daniel J. Powell Jr., PhD, University of Pennsylvania,
8: e60031.
3400 Civic Center Blvd., Bldg. 421, TRC Rm 8-103, Philadelphia, PA
38 Bengsch, B., Ohtani, T., Khan, O., Setty, M., Manne, S., O’Brien, S., Gher- 19104-5156, USA.
ardini, P. F. et al., Epigenomic-guided mass cytometry profiling reveals e-mail: [email protected]
disease-specific features of exhausted CD8 T cells. Immunity. 2018; 48:
1029–1045.e5.

39 Cossarizza, A., Chang, H.-D., Radbruch, A., Acs, A., Adam, D., Adam- Received: 29/4/2021
Klages, S., Agace, W. W. et al., Guidelines for the use of flow cytometry Revised: 7/7/2021
and cell sorting in immunological studies (second edition). Eur. J. Immunol. Accepted: 10/9/2021
2019; 49: 1457–1973. Accepted article online: 10/9/2021

© 2021 Wiley-VCH GmbH www.eji-journal.eu

36
Clinical & Translational Immunology 2020; e1132. doi: 10.1002/cti2.e1132
www.wileyonlinelibrary.com/journal/cti

ORIGINAL ARTICLE

Identification of the immune checkpoint signature of

multiple myeloma using mass cytometry-based
single-cell analysis
Jinheng Wang1 , Yongjiang Zheng2, Chenggong Tu1, Hui Zhang1, Karin Vanderkerken3,
Eline Menu3 & Jinbao Liu1
1
Affiliated Cancer Hospital & Institute of Guangzhou Medical University, Guangzhou Municipal and Guangdong Provincial Key
Laboratory of Protein Modification and Degradation, State Key Laboratory of Respiratory Disease, School of Basic Medical Sciences,
Guangzhou Medical University, Guangzhou, China
2
Department of Hematology, The Third Affiliated Hospital of Sun Yat-Sen University, Guangzhou, China
3
Department of Hematology and Immunology, Myeloma Center Brussels, Vrije Universiteit Brussel, Brussels, Belgium

Correspondence
Abstract
J Wang or J Liu, Affiliated Cancer Hospital &
Institute of Guangzhou Medical University, Objectives. New targets or strategies are needed to increase the
Guangzhou Municipal and Guangdong success of immune checkpoint-based immunotherapy for multiple
Provincial Key Laboratory of Protein
myeloma (MM). However, immune checkpoint signals in MM
Modification and Degradation, State Key
microenvironment have not been fully elucidated. Here, we aimed
Laboratory of Respiratory Disease, School of
Basic Medical Sciences, Guangzhou Medical to have a broad overview of the different immune subsets and
University, 510095 Guangzhou, China. their immune checkpoint status, within the MM
E-mails: [email protected] (JW); microenvironment, and to provide novel immunotherapeutic
[email protected] (JL) targets to treat MM patients. Methods. We performed immune
checkpoint profiling of bone marrow (BM) samples from MM
patients and healthy controls using mass cytometry. With high-
Received 18 January 2020;
dimensional single-cell analysis of 30 immune proteins containing
Revised 5 April 2020;
10 pairs of immune checkpoint axes in 0.55 million of BM cells, an
Accepted 6 April 2020
immune landscape of MM was mapped. Results. We identified an
doi: 10.1002/cti2.e1132 abnormality of immune cell composition by demonstrating a
significant increase in activated CD4 T, CD8 T, CD8+ natural killer
Clinical & Translational Immunology T-like and NK cells in MM BM. Our data suggest a correlation
2020; 9: e1132 between MM cells and immune checkpoint phenotypes and
expand the view of MM immune signatures. Specifically, several
critical immune checkpoints, such as programmed cell death 1 (PD-
1)/PD ligand 2, galectin-9/T-cell immunoglobulin mucin-3, and
inducible T-cell costimulator (ICOS)/ICOS ligand, on both MM and
immune effector cells and a number of activated PD-1+ CD8 T cells
lacking CD28 were distinguished in MM patients. Conclusion. A
clear interaction between MM cells and the surrounding immune
cells was established, leading to immune checkpoint dysregulation.
The analysis of the immune landscape enhances our
understanding of the MM immunological milieu and proposes
novel targets for improving immune checkpoint blockade-based
MM immunotherapy.

ª 2020 The Authors. Clinical & Translational Immunology published by John Wiley & Sons Australia, Ltd on behalf of 2020 | Vol. 9 | e1132
Australian and New Zealand Society for Immunology, Inc.

37
An immune checkpoint signature of multiple myeloma J Wang et al.

Keywords: immune checkpoint, immunotherapy, mass cytometry,

multiple myeloma, single-cell analysis

patients, largely because of the fact that

INTRODUCTION
inhibitory signals inducing the exhaustion and
Multiple myeloma (MM) is a cancer of clonal dysfunction of anticancer immune cells are not
plasma cells preferentially localised in the bone fully and sustainably blocked.10,15 Indeed, as
marrow (BM). The proliferation of MM cells, reported by a phase 1b clinical study, PD-1/PD-L1
together with an MM cell-changed BM axis-based immune checkpoint blockade failed to
microenvironment, suppresses local and systemic control MM progression,16,17 suggesting that this
immunity, eventually leading to an escape from checkpoint may not be the major mediator of
immune surveillance.1 Mechanisms involved in failing anti-MM immunity. Besides PD-1 and CTLA-
MM-induced immunosuppression include 4, many other immune checkpoints have been
dysfunction of T and natural killer (NK) cells,2 discovered and are used for improved immune
disruption of antigen presentation processes,3 checkpoint-based immunotherapy.18 However,
activation of immunosuppressive cells,3,4 immune checkpoint signals in the MM
upregulation of inhibitory immune checkpoints5,6 microenvironment have not been fully elucidated.
and release of immunosuppressive mediators.7 The analysis of immune checkpoints will help us
Comprehensively uncovering the immune status in to better understand the mechanism of immune
the BM microenvironment of MM patients will evasion of MM cells and would allow the
largely facilitate the understanding of the development of potent strategies, focused on the
ongoing process of immunosuppression in MM checkpoint signals that are actually used by MM
progression and therefore promote the cells to evade the immune system.
development of novel immunotherapeutic The most commonly used technique for immune
strategies. phenotyping, flow cytometry, suffers from the
Immunotherapy that involves stimulating and limited detection channels (generally < 15) and
provoking a patients’ own immune system against cumbersome compensation because of spectral
cancer has proven to be very encouraging as overlap, making it difficult to simultaneously
dramatic and durable anticancer responses are detect all immune checkpoint phenotypes. As a
well documented in many cancer types.8,9 cutting-edge single-cell technology, current mass
Blocking inhibitory immune checkpoints on cytometry merging mass spectrometry with flow
immune effector cells results in the reactivation of cytometry permits up to 50 metal isotope tags to
anticancer immunity.10 Immune checkpoints be measured simultaneously on a single cell with
contain a series of costimulatory and coinhibitory minimal/no compensation.19,20 Such high
receptors or ligands expressed on T, NK or multiparametric detection provides an
antigen-presenting cells and mainly function as unprecedented opportunity for deep phenotyping
switches of immune activation or suppression.11 of the tumor immune microenvironment at the
Under normal physiological conditions, immune single-cell level. For now, this powerful innovation
checkpoints maintain self-tolerance and immune has offered insights into the heterogeneity and
homeostasis, whereas malignant cells take complexity of biology and has been used to
advantage of these molecules to achieve immune understand the complex processes in cellular
evasion.12 The most prominent immune development,21 differentiation22 and tumor
checkpoint blocking strategies, such as targeting immunology,23-25 and to explore the potential
cytotoxic T lymphocyte-associated protein 4 immunotherapeutic targets.23
(CTLA-4) and blocking the interaction between In this study, 0.55 million BM cells from 10 MM
programmed cell death 1 (PD-1) and PD ligand 1 patients and five healthy donors (HD) were
(PD-L1), are able to enlist and strengthen the analysed using mass cytometry to elucidate the
immune system to attack cancer cells and have phenotypic diversity and immune checkpoint
achieved clinical success in several cancer types, signature in MM BM ecosystems. Our data reveal
even in metastatic and chemoresistant cancer.13,14 vast phenotypic heterogeneity among both
However, these immunotherapies are unable to malignant and immune cells, identify an
control malignancy in a significant proportion of abnormality of immune cell composition and

2020 | Vol. 9 | e1132 ª 2020 The Authors. Clinical & Translational Immunology published by John Wiley & Sons Australia, Ltd on behalf of
Australian and New Zealand Society for Immunology, Inc.

38
J Wang et al. An immune checkpoint signature of multiple myeloma

suggest links between MM cells and immune expression of immune checkpoint proteins.
checkpoint phenotypes. Through in-depth Important immune checkpoint ligands, including
analyses of 10 pairs of immune checkpoint axes in GAL9, ICOSL, HLA-DR, CD86, PD-L2, and 4-1BBL,
12 identified immune cell types at the single-cell were expressed by more than 10% of MM cells in
level, a picture of the immune checkpoint average (Figure 1f and Supplementary figure 1b).
interaction network that exists in the MM BM We next performed correlation analyses to
microenvironment of these patients was systematically quantify the underlying
established. Several critical immune checkpoints relationships between overall MM burden and
were identified in the MM BM and may serve as MM cells with different immune checkpoint
novel targets for developing more potent and phenotypes. Multiple robust either positive or
efficacious checkpoint blockade-based MM negative relationships were identified (Figure 1g).
immunotherapeutic strategies. Among the positive relationships, GAL9 expression
was most strongly related to MM burden. Also,
the expression of different immune checkpoint
RESULTS
ligands correlated significantly with each other,
such as PD-L2 expression, which correlated with 4-
In-depth immune checkpoint phenotyping
1BBL and CD56 expressions with ICOSL
of MM cells using mass cytometry
(Supplementary figure 1c).
To map the immune checkpoint signatures in the
BM microenvironment of MM patients, we
Immune cell signature in MM BM
implemented a clinical high-dimensional single-
microenvironment
cell profiling study of freshly collected BM from
newly diagnosed and untreated MM patients Next, we used viSNE to visualise the distribution
using mass cytometry. Ten MM BM samples and of the immune cells in the HD and MM BM
five healthy BM samples were included for a samples (equal cell number from each individual)
large-scale mass cytometry analysis (Figure 1a). We and demonstrated a large heterogeneity among
stained prebarcoded BM cells with 30 antibodies MM patients and healthy controls (Figure 2a).
to simultaneously determine the expression of 30 According to the standardised immuno-
markers used to define cell populations and phenotyping for human immunology29 and the
immune checkpoint phenotypes at the single-cell expression of 15 surface markers in HD and MM
level (Figure 1b). As the loss of CD138 caused by BM CD45+ cells displayed on the viSNE map
the cold storage and processing frequently (Figure 2b), 12 major immune cell populations
occurs,26,27 cells with a CD38++CD45/dim were gated on the map (Figure 2c). Natural killer
phenotype were defined as malignant MM cells T (NKT) cells are identified with a CD3+CD56+
(Figure 1c). To comprehensively view the immune phenotype in many studies.30-32 However, only a
checkpoint profile of MM cells from all patients, small proportion of CD3+CD56+ are CD1d-
we generated a single-cell viSNE map to visualise restricted, which is a unique feature of invariant
high-dimensional data in two dimensions.28 This NKT (iNKT) cells. Thus, this population is
analysis demonstrated a clear heterogeneity of frequently referred to as ‘NKT-like’.32 Here, we
MM cells among patients (Figure 1c). On the gated two CD3+CD56+ cell subsets, namely NKT-
viSNE map, clear expression of multiple like and CD8+ NKT-like cells, after excluding CD4,
immunoregulatory proteins, including CTLA-4, CD8 and double-negative (DN) T cells from all
CD56, inducible T-cell costimulator (ICOS), CD3+ cells. As shown by heatmap, the expression
galectin-9 (GAL9), CD86, ICOS ligand (ICOSL), of surface markers in each population was
OX40 and HLA-DR, was observed in different MM identical to the phenotype of indicated immune
cell clusters (Figure 1d). Large proportion of lineages (Figure 2d). After gating on viSNE map,
CD56+ MM cells were detected in 8 of 10 patients, the immune lineages in individual samples were
and GAL9 and ICOSL expressions were widely analysed (Figure 2e), which revealed a
found in MM cells from all patients (Figure 1e), heterogeneity across HD or MM patients.
whereas high PD-L1 or PD-L2 expressions were Although wide variation existed in the
only observed in few MM cells (Supplementary frequencies of each immune cell type in different
figure 1a). These 10 BM samples with 7–41% MM individuals, several significant changes between
cells displayed diverse phenotypes in the HD and MM patients were detected. In the BM of

ª 2020 The Authors. Clinical & Translational Immunology published by John Wiley & Sons Australia, Ltd on behalf of 2020 | Vol. 9 | e1132
Australian and New Zealand Society for Immunology, Inc.

39
An immune checkpoint signature of multiple myeloma J Wang et al.

Figure 1. Characterisation of immune checkpoints of MM cells. (a) The experimental workflow used in this study. (b) Markers used to define
cell populations and immune checkpoint phenotypes. (c) Gating of MM and CD45+ cells (left panel). viSNE map showing 69 253 MM cells from
the BM of MM (n = 10) patients coloured by individual. (d) Cells coloured by normalised expression of indicated immune checkpoint markers on
the viSNE map. (e) A violin plot showing the signal intensity of CD56, GAL9 and ICOSL in MM cells of individual patients. (f) Dot plots showing
the frequency of MM cells among BM cells (left panel) and indicated markers’ positive cells among MM cells for each MM BM sample (right
panel). Dots are coloured by individual. (g) A heatmap showing the Pearson correlation coefficients for relationships between the frequencies of
indicated cell populations. Abs, antibodies; BM, bone marrow; HD, healthy donor; MM, multiple myeloma. MM, n = 10.

MM patients, the proportion of CD4 T, CD8 T, cells rose from 9.49% to 15.36%. Importantly,
CD16+ NK and CD8+ NKT-like cells in CD45+ CD8+ NKT-like cells only accounted for 0.92% of
immune cells was significantly increased along HD BM immune cells in average, whereas it
with the significant decrease in granulocytes, as increased to 4.86% in MM patients (Figure 2f).
compared to those in HD BM cells (Figure 2f). The iNKT cells have been shown to be associated with
average percentage of CD8 T cells increased from MM and are important for antitumor immunity.33
7.77% in HD to 14.82% in MM and that of CD4 T We also examined the proportion of iNKT cells

2020 | Vol. 9 | e1132 ª 2020 The Authors. Clinical & Translational Immunology published by John Wiley & Sons Australia, Ltd on behalf of
Australian and New Zealand Society for Immunology, Inc.

40
J Wang et al. An immune checkpoint signature of multiple myeloma

with the T-cell receptor Va24Ja18 antibody. We the intensity of the expression of these
found that they constitute a minor fraction of BM checkpoints in the corresponding positive cells.
T cells and there is no significant difference in The expression of CD28 was significantly stronger
their percentages between HD and MM patients in CD28+ CD4 T, CD8 T and DNT cells of MM
(Supplementary figure 2a). Moreover, the MM patients. In PD-1+ CD8+ NKT-like and CD8 T cells,
burden was positively correlated with the the PD-1 expression was also significantly
frequency of CD8 T cells in MM patients and increased in MM patients. Many significant
negatively correlated with the frequency of CD16+ changes in the expression of immune checkpoints
NK cells with a trend close to significance in CD4 T, CD8 T, NKT-like or CD8+ NKT-like cells
(Supplementary figure 2b). were discovered (Figure 3g).

The immune checkpoint landscape of MM The immune checkpoint atlas of MM BM

BM T cells non-T cells
To characterise the immune checkpoint The frequencies of PD-1+, PD-L2+, CTLA-4+, ICOS+,
phenotype in MM BM immune cells, we assessed 4-1BBL+, OX40+ and Tim-3+ cells in granulocytes
the expression of all detected immune checkpoint were significantly increased in MM patients,
proteins in CD45+ cells on the viSNE map. ICOSL, although most of them were < 10% (Figure 4a
CD28, CD86 and GAL9 expressions were clearly and b). Granulocytes accounted for the major
observed in several cell subsets (Figure 3a). In provider of ICOSL as more than 85% of them
contrast, no clear accumulative expression of the express ICOSL in both HD and MM patients. The
other immune checkpoints appeared on the viSNE frequencies of PD-1+ and 4-1BB+ cells in undefined
map. However, the normalised mean expression (the rest of) CD45+, Tim-3+ cells in DC, LAG-3+ cells
of these proteins was distinct among the 12 gated in CD16 NK cells, and PD-L2+ cells in CD16+ NK
immune cell populations clustered from viSNE cells were also significantly increased in MM
map (Figure 3b), suggesting the presence of patients (Figure 4c–e and Supplementary figure
heterogeneous subgroups with high immune 5a–c). Although significant differences in the
checkpoint expression in these populations. Thus, percentages of immune checkpoint-positive cells
we first compared the frequencies of immune were not detected in many cell types, the
checkpoint-positive cells in all cell populations of intensity of their expression in several immune cell
HD BM with those of MM patients (Figure 3c and populations was significantly altered in MM
Supplementary figure 3). In BM CD4 T-cell subsets, patients (Figure 4f and g).
the proportions of PD-L1+, PD-L2+, CTLA-4+, 4-1BB+
and 4-1BBL+ cells were consistently < 20%, but
Activation signature of T and NK cells in
were significantly higher in MM patients than
MM BM microenvironment
those in HD. The percentages of CD28+ and ICOS+
CD4 T cells were also significantly higher in MM CD8 T and NK cells are major contributors to
patients than in HD (Figure 3d and Supplementary anticancer immunity and the main targets to be
figure 4a). Moreover, PD-1+, PD-L2+, ICOS+, T-cell reinvigorated by immune checkpoint blockade-
immunoglobulin mucin-3 (Tim-3)+ and lymphocyte based immunotherapy. HLA-DR appears at the
activating 3 (LAG-3)+ CD8 T cells were significantly late stages of activated T and NK cells and has
increased in the BM of MM patients (Figure 3e been widely used as an activation marker.34-36
and Supplementary figure 4b). Additionally, CD38 and HLA-DR are also primarily regarded as
several immune checkpoints were also biomarkers for identifying activated T cells.29,37
significantly increased in other T-cell types, such Here, the activation status of T- and NK cell
as the number of PD-L2+ cells in CD8+ NKT-like subsets was systematically quantified using these
cells; PD-L2+, OX40+ and Tim-3+ cells in NKT-like markers. In the MM BM cells, significant increase
cells; and CTLA-4+ and the number of Tim-3+ cells in activated (HLA-DR+CD38+) cells was found in
in DNT cells. By contrast, some decreases in the CD4 T, CD8 T, NKT-like and CD8+ NKT-like cells
number of immune checkpoint-positive cells were (Figure 5a). Specifically, the average frequency of
observed as well, such as CD28+ and ICOSL+ cells activated cells in CD8 T cells was dramatically
in CD8+ NKT-like cells (Figure 3f and elevated from 11.66% in HD to 40.94% in MM
Supplementary figure 4c–e). We also compared patients (Figure 5b). Similarly, activated NK cells in

ª 2020 The Authors. Clinical & Translational Immunology published by John Wiley & Sons Australia, Ltd on behalf of 2020 | Vol. 9 | e1132
Australian and New Zealand Society for Immunology, Inc.

41
An immune checkpoint signature of multiple myeloma J Wang et al.

Figure 2. Immune cell population changes in the BM of MM patients. (a) A viSNE map displaying gated CD45+ BM cells of five HD and 10 MM
patients coloured by groups. (b) A viSNE map coloured by the normalised expression of indicated markers. (c) A viSNE map coloured by 12 main
cell populations after clustering. (d) A heatmap showing the normalised median expression of 12 indicated markers in 12 cell populations. (e)
Frequencies of 12 cell populations in CD45+ cells for each BM sample. Cell types are indicated by colour. (f) Bar plots showing the frequencies of
indicated populations in BM CD45+ cells of HD and MM patients. HD, n = 5; MM, n = 10. DC, dendritic cells; DNT, double-negative T; Gran,
granulocytes; Mono, monocytes; NK, natural killer; r-CD45+, the rest of CD45+. *P < 0.05 and **P < 0.01.

2020 | Vol. 9 | e1132 ª 2020 The Authors. Clinical & Translational Immunology published by John Wiley & Sons Australia, Ltd on behalf of
Australian and New Zealand Society for Immunology, Inc.

42
J Wang et al. An immune checkpoint signature of multiple myeloma

Figure 3. Immune checkpoint changes in MM BM T cells. (a) A viSNE map coloured by the normalised expression of 15 immune checkpoint
markers. (b) Heatmaps showing the normalised mean expression of 15 immune checkpoint markers in all cell populations (normalised to the
column’s minimum). (c) Contour plots showing the gating strategy and the expression of indicated checkpoint molecules in CD8 T cells of one
representative MM patient. (d, e) Bar plots showing the frequencies of indicated markers’ positive cells in BM (d), CD4 and (e) CD8 T cells of HD
and MM patients. (f) Bar plots showing the significantly changed frequencies of indicated markers’ positive cells in CD8+ NKT-like, NKT-like and
DNT cells of HD and MM patients. (g) Bar plots showing the significantly changed median signal intensity of indicated markers in corresponding
positive T-cell subsets of HD and MM patients. HD, n = 5; MM, n = 10. *P < 0.05 and **P < 0.01.

the BM were also increased in MM patients indicated immune checkpoint protein-expressing

(Figure 5c). A number of strong positive or MM cells in all patients (Supplementary figure 6a).
negative correlations were revealed between the In activated (HLA-DR+) T- or NK cell subsets,
frequencies of activated T- or NK cell subsets and several changes in immune checkpoint phenotype

ª 2020 The Authors. Clinical & Translational Immunology published by John Wiley & Sons Australia, Ltd on behalf of 2020 | Vol. 9 | e1132
Australian and New Zealand Society for Immunology, Inc.

43
An immune checkpoint signature of multiple myeloma J Wang et al.

Figure 4. Immune checkpoint changes in MM BM non-T immune cells. (a) Contour plots showing the gating strategy and the expression of
indicated checkpoint molecules in granulocytes of one representative MM patient. (b–e) Bar plots showing the frequencies of indicated markers’
positive cells in BM (b) granulocytes, (c) DC, (d) CD16 NK and (e) CD16+ NK cells of HD and MM patients. (f, g) Bar plots showing the
significantly changed median signal intensity of indicated markers in corresponding positive (f) granulocytes, B and r-CD45+, and (g) DC, CD16+
and CD16 NK cells of HD and MM patients. HD, n = 5; MM, n = 10. *P < 0.05 and **P < 0.01.

appeared in MM patients compared with those in two markers to examine whether increased CD8 T
HD (Figure 5d–f), changes were also found in cells in the BM are specific against MM cells.
inactivated (HLA-DR) T cells (Supplementary However, above 90% of CD8 T or activated
figure 6b and c). In addition, the expression of CD8 T cells are CD39– and CD103–negative
important immune checkpoints, including PD-1, (Supplementary figure 6d), suggesting that
CD28 and ICOS, was changed in activated (HLA- bystander T cells instead of tumor-specific CD8 T
DR+) CD4 and CD8 T cells (Figure 5g). cells are abundant in MM BM. To identify the
Coexpression of CD39 and CD103 has been used immune checkpoint phenotypes in activated cells,
to identify the tumor-specific CD8+ T cells in we compared the frequencies of the immune
human tumors.38,39 Here, we introduced these marker-expressing cells in inactivated with

2020 | Vol. 9 | e1132 ª 2020 The Authors. Clinical & Translational Immunology published by John Wiley & Sons Australia, Ltd on behalf of
Australian and New Zealand Society for Immunology, Inc.

44
J Wang et al. An immune checkpoint signature of multiple myeloma

activated T or NK cells. Activated CD4 T cells 89 (Figure 6d). In addition, MM burden was
expressed more Tim-3, PD-1, GAL9, CTLA-4, ICOS significantly correlated with the frequencies of
and 4-1BB than inactivated cells in both HD and clusters 32, 48, 76, 82, 92 and 96 in MM patients
MM patients. More activated CD8 T cells (Figure 6e), indicating that the changes in these T-
expressed PD-1, GAL9, ICOS, CTLA-4 and Tim-3 cell clusters are MM cell-dependent.
(Figure 5h). Moreover, compared with inactivated
cells, more activated NKT-like cells expressed
Immune checkpoint network in the MM BM
CTLA-4; more activated CD16 NK cells expressed
microenvironment
CTLA-4, Tim-3 and GAL9; and more activated
CD16+ NK cells expressed 4-1BB (Supplementary We summarised the top 3 cell types providing
figure 6e). immune checkpoint-related receptors or ligands in
MM patients (Figure 7a). Based on these main
providers and the expression of immune
In-depth and systematic analyses of the
checkpoint molecules in MM cells, a list and a
immune checkpoint profile of MM BM T
network describing the interactions among MM
cells
and immune cells through immune checkpoints
As T cells are the primary anticancer contributor, were established (Figure 7b and c). Considering
we next systematically analysed the immune the large heterogeneity among MM patients, we
checkpoint phenotype of all possible exclusively also built an immune checkpoint network for
and significantly changed T-cell clusters. From the each MM patient (Supplementary figure 8).
viSNE containing all CD3 T cells, we observed a
huge heterogeneity of the T-cell compartments,
DISCUSSION
regarding the expression of immune modulatory
proteins (Supplementary figure 7a). We next The BM contains a complex environment and is
introduced spanning-tree progression analysis of filled with numerous kinds of immunoregulatory
density-normalised events (SPADE) analysis40 to signal from both immune and non-immune cells.
divide all T cells into 100 minor clusters (nodes) In the MM BM microenvironment, non-immune
containing cells with similar phenotypes. On the cells, such as stromal cells, regulate
SPADE tree, we were able to characterise the immunosuppression through cell-to-cell contact
immune checkpoint phenotype of each cluster and release extracellular vesicles and thus favor
and clearly observe the differences in these immune evasion of MM cells.41 Immune
clusters in each individual (Figure 6a and checkpoints expressed on immune cells maintain
Supplementary figure 7b). Using cytoClusterR, the the immune homeostasis, whereas MM cells
heterogeneity of immune checkpoint receptor enhance the suppression signal to escape from
signatures across 100 T-cell clusters from all 10 immune surveillance. Immune checkpoint
MM patients or five HD was obviously revealed on blockade can break this malignant cell-induced
heatmaps (Figure 6b). Clusters 82, 92, 89, 68 and inhibitory communication and thus lead to the
42 were specifically presented in MM patients. In reinvigoration of anticancer immunity. Success of
each cluster, different median expressions of immune checkpoint therapies largely relies on the
immune checkpoint protein are summarised targets responsible for cancer-induced immune
(Figure 6b). Among these 100 T-cell clusters, the suppression. To improve our understanding of the
frequencies of 42 clusters in MM patients were immune signature and immune checkpoint
significantly different from those in HD (Figure 6c abnormalities in the MM BM microenvironment,
and Supplementary figure 7c). Twenty-eight we performed a high-dimensional single-cell
clusters displayed an activated phenotype (HLA- analysis of the immune checkpoint molecules in
DR+) and were significantly increased in MM healthy and MM BM samples. This high-quality
patients (Figure 6c), indicating that these T-cell data set identifies an unambiguous immune
clusters may play pivotal roles in remodelling the checkpoint network in the MM immunologic
MM BM immune microenvironment. Among these milieu of these patients (Figure 7b and c) and
28 clusters, eight CD8 T-cell clusters, including establishes a powerful new level of insights into
clusters 37, 32, 39, 21, 73, 89, 68 and 42, were MM checkpoint immunotherapy.
activated and PD-1+, whereas all these clusters Mass cytometry has been recently used to
were deficient in CD28 expression, except cluster identify T-cell heterogeneity and early alterations

ª 2020 The Authors. Clinical & Translational Immunology published by John Wiley & Sons Australia, Ltd on behalf of 2020 | Vol. 9 | e1132
Australian and New Zealand Society for Immunology, Inc.

45
An immune checkpoint signature of multiple myeloma J Wang et al.

Figure 5. Changes in T-cell activation status in the BM of MM patients. (a) Contour plots showing the expression of CD38 and HLA-DR in 4 T-
cell subsets of one representative HD or MM patient. (b) Bar plots showing the frequencies of indicated cell clusters in BM T-cell subsets of HD
and MM patients. (c) Contour plots showing the expression of HLA-DR in NK cell subsets of one representative HD or MM patient (left panel).
Bar plots showing the frequencies of indicated clusters in BM NK cell subsets of HD and MM patients (right panel). (d) Bar plots showing the
significantly changed frequencies of indicated markers’ positive cells in HLA-DR+CD38+ T-cell subsets of HD and MM patients. (e) Bar plots
showing the significantly changed frequencies of indicated markers’ positive cells in HLA-DR+CD38 T-cell subsets of HD and MM patients. (f)
Bar plots showing the significantly changed frequencies of indicated markers’ positive cells in HLA-DR+ NK cell subsets of HD and MM patients.
(g) Bar plots showing the significantly changed median signal intensity of indicated markers in corresponding positive HLA-DR+CD38+ CD4 T,
HLA-DR+CD38 CD4 T and HLA-DR+CD38+ CD8 T cells of HD and MM patients. (h) Dot plots showing the significantly changed frequencies of
the indicated markers’ positive cells in HLA-DRCD38 and HLA-DR+CD38+ T-cell subsets of the individual. HD, n = 5; MM, n = 10. *P < 0.05,
**P < 0.01, ***P < 0.001 and ****P < 0.0001.

in resident T cells, and innate and myeloid cells in from dysproteinaemia patients, including MGUS,
the BM of MM.42,43 Kourelis et al.42 have MM and AL amyloidosis, at diagnosis and after
evaluated 33 immune markers, including five chemotherapy, and autologous stem cell
immune checkpoint molecules, in BM samples transplant using mass cytometry. Similar to our

2020 | Vol. 9 | e1132 ª 2020 The Authors. Clinical & Translational Immunology published by John Wiley & Sons Australia, Ltd on behalf of
Australian and New Zealand Society for Immunology, Inc.

46
J Wang et al. An immune checkpoint signature of multiple myeloma

Figure 6. Identification of the immune checkpoint signature of T cell in MM patients. (a) A SPADE tree describing 100 small T-cell clusters of
one representative HD or MM patient coloured by the median expression of indicated markers. T-cell subpopulations are gated with a grey
colour, and PD-1+ subsets are gated with a deep grey area. (b) Heatmaps showing the normalised median expression of indicated markers in 100
small T-cell clusters of all MM patients and all HD and displaying the differences in markers’ expression between T-cell clusters of MM patients
and HD (right panel). (c) Bar plots showing the significantly changed frequencies of T-cell clusters (nodes) of HD and MM patients. (d)
A heatmap showing the normalised median expression of indicated markers in significantly changed HLA-R+ T-cell clusters of MM patients. Red
boxes indicate PD-1+HLA-DR+CD38+ CD8 T-cell clusters. (e) Dot plots showing the Pearson correlation coefficients for relationships between the
frequencies of MM cells and indicated T-cell clusters. HD, n = 5; MM, n = 10. *P < 0.05, **P < 0.01 and ***P < 0.001.

results, they also found a very low level of CTLA-4 DC, express very low level of PD-L1, further
in both CD4 and CD8 T cells and that PD-1 is confirming the lack of PD-1/PD-L1 checkpoint
expressed by several T-cell clusters, but not by all signalling. The other recent study also analysed
T cells. All identified BM cell types, except myeloid the BM T cells from 7 HD and 10 MM patients and

ª 2020 The Authors. Clinical & Translational Immunology published by John Wiley & Sons Australia, Ltd on behalf of 2020 | Vol. 9 | e1132
Australian and New Zealand Society for Immunology, Inc.

47
An immune checkpoint signature of multiple myeloma J Wang et al.

Figure 7. The immune checkpoint network in the MM BM microenvironment. (a) Dot plots showing the top 3 frequencies of indicated markers’
positive cells in immune cell types. Dots are coloured by individual. (b) A table listing all the important checkpoints and their top 3 or 4 providers.
(c) A schematic diagram showing the main provider cells of immune checkpoint ligands and receptors, and the network among them in the MM
BM microenvironment. Act, activated. MM, n = 10.

the BM myeloid cells from 4 HD and 8 MM blocking strategy has also become a focus of MM
patients using mass cytometry.43 They discovered immunotherapy and plenty of clinical trials are
greater terminal effector differentiation in conducted.44 However, single-agent therapy with
memory T cells and an increased PD-L1 expression PD-1 inhibitors fails to induce significant clinical
on myeloid cells from MM patients than healthy responses in a phase 1b study,16 suggesting that
donors. However, detailed status of immune PD-1 blockade alone is insufficient to reinvigorate
checkpoints, as well as the cell types providing a clinically meaningful anti-MM immunity.
checkpoint signals, has not been identified in Discrepant results concerning PD-L1 expression on
these previous studies. Here, we devoted to MM cells have been reported.45 Several studies
systemically delineate the immune checkpoint have confirmed the limited expression of PD-L1 on
signature of MM by measuring 10 pairs of MM cells46-48; in contrast, higher PD-L1 has been
immune checkpoint axes in freshly isolated BM also found in MM cells than plasma cells from
samples from MM patients without treatment and HD.5,49 Our comprehensive data revealed a low
our data would maximally reflect the real immune frequency (< 12%) of PD-L1 expression in MM cells
status of MM BM microenvironment. from all 10 MM patients. However, the expression
Malignant cells offer a variety of immune of PD-L2, another ligand for PD-1, on MM cells
checkpoint ligands to match receptors on immune was relatively higher than PD-L1. Anyhow, ligands
cells and thus regulate anticancer immunity. With of PD-1 were not widely expressed by MM cells,
the successful application of PD-1/PD-L1 axis implicating the existence of other possible
inhibitors in solid tumor immunotherapy, this participants in inhibitory immunity. We validated

2020 | Vol. 9 | e1132 ª 2020 The Authors. Clinical & Translational Immunology published by John Wiley & Sons Australia, Ltd on behalf of
Australian and New Zealand Society for Immunology, Inc.

48
J Wang et al. An immune checkpoint signature of multiple myeloma

here that several immune checkpoint ligands, Deep analysis of T-cell profiling identified several
including GAL9, ICOSL, HLA-DR, CD86, PD-L2 and specifically activated CD8 T-cell clusters highly
4-1BBL, were more generally presented on MM expressing PD-1 in MM patients, whereas most of
cells and these ligands are able to largely influence them are deficient in CD28 expression, a critical T-
the immune response through binding to their cell costimulatory receptor that binds to B7
receptors on immune effector cells. molecules, including CD80 and CD86.61 The failure
A significant positive correlation between MM of PD-1 inhibitors in MM immunotherapy may
burden and GAL9 expression, together with the result from the deficiency of CD28 in activated CD8+
high frequency of GAL9 expression on MM cells, T cells as substantial evidences have demonstrated
emphasises the possible contribution of this that successful reinvigoration of exhausted CD8+ T
ligand to the MM immune microenvironment. In cells by PD-1/PD-L1 blockade is dependent on
addition, Tim-3, a receptor of GAL9,50 was CD28.62,63 Most likely, once CD28 signalling is
expressed by activated CD8 T, NKT-like, DNT cells restored in these increased numbers of activated
and DC in MM patients. Tim-3-GAL9 axis provides CD8+ T cells, strong anti-MM immunity will be
inhibitory immune signals to activated T cells,51 achieved for controlling MM growth.
and immunotherapy targeting Tim-3 and PD-1 New targets or strategies are needed to increase
pathways enables the reversion of T-cell the success of immune checkpoint-based
exhaustion and restoration of antitumor immunotherapy for MM. By fine-grained analysis
immunity,52 thus suggesting a possible use of this of the immune cells in the MM BM
strategy to reconstruct anti-MM immunity. microenvironment, this study provides a detailed
ICOSL was also expressed by most of MM cells, and atlas of the infiltrating immune cells in MM,
its receptor ICOS was increasingly detected in identifies immune checkpoints change that are
20–40% of CD4 or CD8 T cells of MM patients. Being unique to the MM immunologic milieu, and reveals
in line with this mechanism, a higher percentage of distinct immune subsets that may be responsible
ICOS+ cells in follicular helper T cells has been found for anti-MM immunosuppression. These data will
in MM patients than healthy controls.53 The ICOS/ be a valuable resource for future research to
ICOSL signal can mediate helper T-cell immunity and explore more efficient immunotherapy strategies
regulate effector T-cell differentiation.54 In vitro, tailored to restore anti-MM immunity through
ICOS/ICOSL blockade significantly reduced the inhibition of immune checkpoints. The large
generation of MM cell-induced inhibitory CD4+ Treg individual heterogeneity in immune checkpoint
cells,55,56 and lenalidomide, a clinically approved networks among MM patients also emphasises the
anti-MM immunomodulatory drug, could inhibit necessity of personalised strategies for a successful
ICOSL expression in MM cells57 and enhance PD-1/ MM immunotherapy. Our findings demonstrating
PD-L1 blockade-induced anticancer immunity in MM several potential immune checkpoint targets
patients.58 These evidences, together with our warrant further functional investigation into
results, underline ICOS/ICOSL blockade as a possible developing novel strategies for MM
enhancer for anti-MM immunotherapeutic immunotherapy. In addition, non-immune cell
strategies. components, such as stromal cells and extracellular
T and NK cells are at the forefront of anticancer vesicles, which also play an important role in
immune responses, and quantitative and functional regulating immunosuppression in the MM BM, also
abnormalities in these cells’ subsets have been well need to be taken into account in discovering novel
identified in the MM BM microenvironment.2,59,60 targets for MM treatment in future.
The discovery of significant increases in CD4 T, CD8
T, CD16+ NK and CD8+ NKT-like cells in MM BM
compared with HD BM confirms an abnormal METHODS
immune cell composition induced by MM cells.
Remarkably, these increased T or NK cells are Human specimens
activated in the MM samples, but with a
suppressive phenotype as several inhibitory Multiple myeloma BM samples were collected from MM
patients undergoing BM biopsy for diagnosis, and healthy
receptors, such as PD-1 and Tim-3, were increased. BM samples were obtained from donors undergoing BM
Because of the fact that CTLA-4, 4-1BB and LAG-3 biopsy for BM donation. Informed consents in accordance
were expressed only by very few CD8 T cells, with the Declaration of Helsinki were obtained from all
targeting those checkpoints might be less effective. participants. All participants were recruited at the Third

ª 2020 The Authors. Clinical & Translational Immunology published by John Wiley & Sons Australia, Ltd on behalf of 2020 | Vol. 9 | e1132
Australian and New Zealand Society for Immunology, Inc.

49
An immune checkpoint signature of multiple myeloma J Wang et al.

Affiliated Hospital of Sun Yat-sen University. All protocols

were reviewed and approved by the Hospital Ethics Data analysis
Committee. The clinical characteristics of all participants are
Individual debarcoded files were uploaded to an online
listed in Supplementary table 1.
single-cell analyser, Cytobank (Beckman Coulter, Brea, CA,
USA).65 Beads, debris and doublets were excluded from the
events, and the single-cell data were subsequently used for
Sample processing
high-dimensional analyses. Contour plots, viSNE, SPADE and
Bone marrow samples were collected into sodium heparin heatmaps were implemented using Cytobank. The
tubes. To maximally maintain the immune profile, freshly frequency of positive cells in each gated population was
isolated BM cells were directly fixed using an optimised and determined using FlowJo (FlowJo LLC, Ashland, OR, USA).
well-established fixing method with minimal effects on Bar plot, violin plot and heatmap of correlationship were
target epitope.64 About 1–2 mL of BM samples was fixed generated using the GraphPad Prism software (GraphPad,
with Fix I Buffer (Fluidigm, South San Francisco, CA, USA) for San Diego, CA, USA). The comparison of HD and MM
10 min at RT, and red blood lysis buffer was used to fully SPADE node was implemented using the cytoClustR R
remove the erythrocytes. Cells were then resuspended in cell package developed in Kordasti Lab from King’s College
staining buffer (CSB) and Dulbecco’s phosphate-buffered London. SPSS 20.0 software (IBM, Armonk, NY, USA) was
saline, supplemented with 0.5% bovine serum albumin and used for the Pearson correlation analyses.
0.02% sodium azide, containing 10% dimethyl sulphoxide,
and stored at 80°C until cell staining was performed.
Statistical analysis
The Mann–Whitney U-test was used to determine the
Barcoding
statistical significance between the two groups. A paired t-
To eliminate sample-specific staining variation, all samples test was performed on the frequencies of different cell
were barcoded first and then stained, processed and subsets in individuals. Error bars represent mean standard
acquired as one multiplexed sample. A total of 0.5 9 106 error of mean (sem). A P-value < 0.05 was considered as
fixed cells from each samples were washed thrice with CSB statistically significant.
and washed twice with 19 Barcode Perm Buffer (Fluidigm).
These samples were then barcoded using a 20-Plex Pd
Barcoding Kit (Fluidigm). Each sample was washed thrice ACKNOWLEDGMENTS
with CSB after incubation with different barcodes for
This work was supported by the National Natural Science
30 min at RT, and all samples were combined together into
Foundation of China (Grant No. 81700203), the Guangdong
one tube for antibody staining.
Basic and Applied Basic Research Foundation
(2019A1515011126), the National Funds for Developing Local
Colleges and Universities (Grant No. B16056001) and the
Antibody staining
Natural Science Foundation Research Team of Guangdong
Combined samples were washed once with CSB and Province (Grant No. 2018B030312001). We thank all patients
incubated with Human Fc Receptor Binding Inhibitor or healthy donors for supporting this work.
Antibody (Thermo Fisher, Waltham, MA, USA) for 10 min at
RT to lower non-specific binding. Anti-human ICOSL-biotin
(BioLegend, San Diego, CA, USA) was added to the samples CONFLICT OF INTEREST
for incubation for another 30 min at RT. These cells were
The authors declare no conflict of interest.
washed twice with CSB and stained with 29 metal isotope-
tagged antibodies and 1 metal-labelled antibody against
biotin (Supplementary table 2) for 30 min at RT. These
stained cells were washed thrice with CSB and incubated
AUTHORS’ CONTRIBUTION
with 1 mL Fix & Perm Buffer (Fluidigm) containing 125 nM JW and JL conceived the idea and supervised the
Intercalator-Ir (Fluidigm) overnight at 4°C. experiments. YZ obtained the informed consents and
collected samples. JW, CT and HZ implemented the
experiments. JW analysed the data and wrote the
CyTOF data acquisition manuscript. KV, EM and JL provided critical suggestions and
revised the manuscript. All authors read and approved the
Samples were washed twice with CSB and twice with final manuscript.
ultrapure water. Immediately prior to data acquisition, the
sample was resuspended in ultrapure water containing 15%
EQ Four Element Calibration Beads (Fluidigm) and filtered Data availability statement
through a 38-lm cell strainer. The sample was acquired on
a Helios mass cytometer (Fluidigm) at an acquisition rate of Mass cytometry data that support the findings of this study
< 500 events/s. Bead-based normalisation and debarcoding are deposited in the FlowRepository database (No. FR-FCM-
were completed using CyTOF software 6.7 (Fluidigm). Z29D).

2020 | Vol. 9 | e1132 ª 2020 The Authors. Clinical & Translational Immunology published by John Wiley & Sons Australia, Ltd on behalf of
Australian and New Zealand Society for Immunology, Inc.

50
J Wang et al. An immune checkpoint signature of multiple myeloma

REFERENCES malignancy: preliminary results of a phase Ib study. J

Clin Oncol 2016; 34: 2698–2704.
1. Wang J, Hendrix A, Hernot S et al. Bone marrow 18. Pardoll DM. The blockade of immune checkpoints in
stromal cell-derived exosomes as communicators in drug cancer immunotherapy. Nat Rev Cancer 2012; 12: 252–
resistance in multiple myeloma cells. Blood 2014; 124: 264.
555–566. 19. Bendall SC, Simonds EF, Qiu P et al. Single-cell mass
2. Dosani T, Carlsten M, Maric I, Landgren O. The cellular cytometry of differential immune and drug responses
immune system in myelomagenesis: NK cells and T cells across a human hematopoietic continuum. Science
in the development of myeloma and their uses in 2011; 332: 687–696.
immunotherapies. Blood Cancer J 2015; 5: e306. 20. Bodenmiller B, Zunder ER, Finck R et al. Multiplexed
3. Ghobrial I, Cruz CH, Garfall A et al. Immunotherapy in mass cytometry profiling of cellular states perturbed by
multiple myeloma: accelerating on the path to the small-molecule regulators. Nat Biotechnol 2012; 30:
patient. Clin Lymphoma Myeloma Leuk 2019; 19: 332–344. 858–867.
4. Ramachandran IR, Martner A, Pisklakova A et al. 21. Bendall SC, Davis KL, el Amir AD et al. Single-cell
Myeloid-derived suppressor cells regulate growth of trajectory detection uncovers progression and
multiple myeloma by inhibiting T cells in bone marrow. regulatory coordination in human B cell development.
J Immunol 2013; 190: 3815–3823. Cell 2014; 157: 714–725.
5. Liu J, Hamrouni A, Wolowiec D et al. Plasma cells from 22. Porpiglia E, Samusik N, Ho ATV et al. High-resolution
multiple myeloma patients express B7–H1 (PD-L1) and myogenic lineage mapping by single-cell mass
increase expression after stimulation with IFN-c and TLR cytometry. Nat Cell Biol 2017; 19: 558–567.
ligands via a MyD88-, TRAF6-, and MEK-dependent 23. Lavin Y, Kobayashi S, Leader A et al. Innate immune
pathway. Blood 2007; 110: 296–304. landscape in early lung adenocarcinoma by paired
6. Lucas F, Pennell M, Huang Y et al. T cell transcriptional single-cell analyses. Cell 2017; 169: 750–765.e17.
profiling and immunophenotyping uncover LAG3 as a 24. Chevrier S, Levine JH, Zanotelli VRT et al. An immune
potential significant target of immune modulation in atlas of clear cell renal cell carcinoma. Cell 2017; 169:
multiple myeloma. Biol Blood Marrow Transplant 2020; 736–749.e18.
26: 7–15. 25. Wagner J, Rapsomaniki MA, Chevrier S et al. A single-
7. Wang J, De Veirman K, Faict S et al. Multiple myeloma cell atlas of the tumor and immune ecosystem of
exosomes establish a favourable bone marrow human breast cancer. Cell 2019; 177: 1330–1345.e18.
microenvironment with enhanced angiogenesis and 26. Lin P, Owens R, Tricot G, Wilson CS. Flow cytometric
immunosuppression. J Pathol 2016; 239: 162–173. immunophenotypic analysis of 306 cases of multiple
8. Jelinek T, Mihalyova J, Kascak M, Duras J, Hajek R. PD-1/ myeloma. Am J Clin Pathol 2004; 121: 482–488.
PD-L1 inhibitors in haematological malignancies: update 27. Reid S, Yang S, Brown R et al. Characterisation and
2017. Immunology 2017; 152: 357–371. relevance of CD138-negative plasma cells in plasma cell
9. Minn AJ, Wherry EJ. Combination cancer therapies with myeloma. Int J Lab Hematol 2010; 32: e190–e196.
immune checkpoint blockade: convergence on 28. el Amir AD, Davis KL, Tadmor MD et al. viSNE enables
interferon signaling. Cell 2016; 165: 272–275. visualization of high dimensional single-cell data and
10. Pitt JM, Vetizou M, Daillere R et al. Resistance reveals phenotypic heterogeneity of leukemia. Nat
mechanisms to immune-checkpoint blockade in cancer: Biotechnol 2013; 31: 545–552.
tumor-intrinsic and -extrinsic factors. Immunity 2016; 29. Maecker HT, McCoy JP, Nussenblatt R. Standardizing
44: 1255–1269. immunophenotyping for the human immunology
11. Syn NL, Teng MWL, Mok TSK, Soo RA. De-novo and project. Nat Rev Immunol 2012; 12: 191–200.
acquired resistance to immune checkpoint targeting. 30. Chan WK, Rujkijyanont P, Neale G et al. Multiplex and
Lancet Oncol 2017; 18: e731–e741. genome-wide analyses reveal distinctive properties of
12. Topalian SL, Taube JM, Anders RA, Pardoll DM. KIR+ and CD56+ T cells in human blood. J Immunol
Mechanism-driven biomarkers to guide immune 2013; 191: 1625–1636.
checkpoint blockade in cancer therapy. Nat Rev Cancer 31. Campbell JJ, Qin S, Unutmaz D et al. Unique
2016; 16: 275–287. subpopulations of CD56+ NK and NK-T peripheral blood
13. Wilson RAM, Evans TRJ, Fraser AR, Nibbs RJB. Immune lymphocytes identified by chemokine receptor
checkpoint inhibitors: new strategies to checkmate expression repertoire. J Immunol 2001; 166: 6477–6482.
cancer. Clin Exp Immunol 2018; 191: 133–148. 32. Krijgsman D, Hokland M, Kuppen PJK. The role of
14. Brunner-Weinzierl MC, Rudd CE. CTLA-4 and PD-1 natural killer T cells in cancer-a phenotypical and
control of T-cell motility and migration: implications for functional approach. Front Immunol 2018; 9: 367.
tumor immunotherapy. Front Immunol 2018; 9: 2737. 33. Tiwary S, Berzofsky JA, Terabe M. Altered lipid tumor
15. Jenkins RW, Barbie DA, Flaherty KT. Mechanisms of environment and its potential effects on NKT cell function
resistance to immune checkpoint inhibitors. Br J Cancer in tumor immunity. Front Immunol 2019; 10: 2187.
2018; 118: 9–16. 34. Arruvito L, Payaslian F, Baz P et al. Identification and
16. Suen H, Brown R, Yang S et al. The failure of immune clinical relevance of naturally occurring human CD8+HLA-
checkpoint blockade in multiple myeloma with PD-1 DR+ regulatory T cells. J Immunol 2014; 193: 4469–4476.
inhibitors in a phase 1 study. Leukemia 2015; 29: 1621– 35. Evans JH, Horowitz A, Mehrabi M et al. A distinct
1622. subset of human NK cells expressing HLA-DR expand in
17. Lesokhin AM, Ansell SM, Armand P et al. Nivolumab in response to IL-2 and can aid immune responses to BCG.
patients with relapsed or refractory hematologic Eur J Immunol 2011; 41: 1924–1933.

ª 2020 The Authors. Clinical & Translational Immunology published by John Wiley & Sons Australia, Ltd on behalf of 2020 | Vol. 9 | e1132
Australian and New Zealand Society for Immunology, Inc.

51
An immune checkpoint signature of multiple myeloma J Wang et al.

36. Christopoulos P, Pfeifer D, Bartholome K et al. 53. Zhou DM, Xu YX, Zhang LY et al. The role of follicular
Definition and characterization of the systemic T-cell T helper cells in patients with malignant lymphoid
dysregulation in untreated indolent B-cell lymphoma disease. Hematology 2017; 22: 412–418.
and very early CLL. Blood 2011; 117: 3836–3846. 54. Wikenheiser DJ, Stumhofer JS. ICOS co-stimulation:
37. McElroy AK, Akondy RS, Davis CW et al. Human Ebola friend or foe? Front Immunol 2016; 7: 304.
virus infection results in substantial immune activation. 55. Feyler S, Scott GB, Parrish C et al. Tumour cell
Proc Natl Acad Sci USA 2015; 112: 4719–4724. generation of inducible regulatory T-cells in multiple
38. Simoni Y, Becht E, Fehlings M et al. Bystander CD8+ T myeloma is contact-dependent and antigen-presenting
cells are abundant and phenotypically distinct in cell-independent. PLoS One 2012; 7: e35981.
human tumour infiltrates. Nature 2018; 557: 575–579. 56. Raja KRM, Hajek R. Contribution of regulatory T cells
39. Duhen T, Duhen R, Montler R et al. Co-expression of to immunosuppression and disease progression in
CD39 and CD103 identifies tumor-reactive CD8 T cells in multiple myeloma patients. Oncoimmunology 2013; 2:
human solid tumors. Nat Commun 2018; 9: 2724. e25619.
40. Qiu P, Simonds EF, Bendall SC et al. Extracting a cellular 57. Scott GB, Carter C, Parrish C, Wood PM, Cook G.
hierarchy from high-dimensional cytometry data with Downregulation of myeloma-induced ICOS-L and
SPADE. Nat Biotechnol 2011; 29: 886–891. regulatory T cell generation by lenalidomide and
41. Wang J, De Veirman K, De Beule N et al. The bone dexamethasone therapy. Cell Immunol 2015; 297: 1–9.
marrow microenvironment enhances multiple myeloma 58. Gorgun G, Samur MK, Cowens KB et al. Lenalidomide
progression by exosome-mediated activation of enhances immune checkpoint blockade-induced
myeloid-derived suppressor cells. Oncotarget 2015; 6: immune response in multiple myeloma. Clin Cancer Res
43992–44004. 2015; 21: 4607–4618.
42. Kourelis TV, Villasboas JC, Jessen E et al. Mass 59. Perez-Andres M, Almeida J, Martin-Ayuso M et al.
cytometry dissects T cell heterogeneity in the immune Characterization of bone marrow T cells in monoclonal
tumor microenvironment of common dysproteinemias gammopathy of undetermined significance, multiple
at diagnosis and after first line therapies. Blood Cancer myeloma, and plasma cell leukemia demonstrates
J 2019; 9: 72. increased infiltration by cytotoxic/Th1 T cells
43. Bailur JK, McCachren SS, Doxie DB et al. Early demonstrating a squed TCR-Vb repertoire. Cancer 2006;
alterations in stem-like/resident T cells, innate and 106: 1296–1305.
myeloid cells in the bone marrow in preneoplastic 60. Tamura H. Immunopathogenesis and immunotherapy
gammopathy. JCI Insight 2019; 5: 127807. of multiple myeloma. Int J Hematol 2018; 107: 278–285.
44. Lesokhin AM, Bal S, Badros AZ. Lessons learned from 61. Esensten JH, Helou YA, Chopra G, Weiss A, Bluestone
checkpoint blockade targeting PD-1 in multiple JA. CD28 costimulation: from mechanism to therapy.
myeloma. Cancer Immunol Res 2019; 7: 1224–1229. Immunity 2016; 44: 973–988.
45. Jelinek T, Paiva B, Hajek R. Update on PD-1/PD-L1 inhibitors 62. Kamphorst AO, Wieland A, Nasti T et al. Rescue of
in multiple myeloma. Front Immunol 2018; 9: 2431. exhausted CD8 T cells by PD-1-targeted therapies is
46. Paiva B, Azpilikueta A, Puig N et al. PD-L1/PD-1 CD28-dependent. Science 2017; 355: 1423–1427.
presence in the tumor microenvironment and activity of 63. Hui EF, Cheung J, Zhu J et al. T cell costimulatory
PD-1 blockade in multiple myeloma. Leukemia 2015; 29: receptor CD28 is a primary target for PD-1-mediated
2110–2113. inhibition. Science 2017; 355: 1428–1433.
47. Kelly KR, Espitia CM, Zhao W et al. Oncolytic reovirus 64. Krutzik PO, Clutter MR, Nolan GP. Coordinate analysis
sensitizes multiple myeloma cells to anti-PD-L1 therapy. of murine immune cell surface markers and intracellular
Leukemia 2018; 32: 230–233. phosphoproteins by flow cytometry. J Immunol 2005;
48. Favreau M, Venken K, Faict S et al. Both mucosal- 175: 2357–2365.
associated invariant and natural killer T-cell deficiency 65. Kotecha N, Krutzik PO, Irish JM. Web-based analysis
in multiple myeloma can be countered by PD-1 and publication of flow cytometry experiments. Curr
inhibition. Haematologica 2017; 102: e266–e270. Protoc Cytom 2010; Chapter 10: Unit10.17.
49. Tamura H, Ishibashi M, Yamashita T et al. Marrow
stromal cells induce B7–H1 expression on myeloma cells,
generating aggressive characteristics in multiple Supporting Information
myeloma. Leukemia 2013; 27: 464–472.
50. Zhu C, Anderson AC, Schubart A et al. The Tim-3 ligand Additional supporting information may be found
galectin-9 negatively regulates T helper type 1 online in the Supporting Information section at
immunity. Nat Immunol 2005; 6: 1245–1252. the end of the article.
51. Perez-Gracia JL, Labiano S, Rodriguez-Ruiz ME,
Sanmamed MF, Melero I. Orchestrating immune check-
point blockade for cancer immunotherapy in
This is an open access article under the terms of
combinations. Curr Opin Immunol 2014; 27: 89–97.
the Creative Commons Attribution License,
52. Sakuishi K, Apetoh L, Sullivan JM et al. Targeting Tim-3 which permits use, distribution and reproduction
and PD-1 pathways to reverse T cell exhaustion and restore in any medium, provided the original work is
anti-tumor immunity. J Exp Med 2010; 207: 2187–2194. properly cited.

2020 | Vol. 9 | e1132 ª 2020 The Authors. Clinical & Translational Immunology published by John Wiley & Sons Australia, Ltd on behalf of
Australian and New Zealand Society for Immunology, Inc.

52
Original Article

A Phase 1b/2 Study of Azacitidine With PD-L1 Antibody

Avelumab in Relapsed/Refractory Acute Myeloid Leukemia
1
Kapil Saxena, MD ; Shelley M. Herbrich, PhD1; Naveen Pemmaraju, MD 1
; Tapan M. Kadia, MD 1
;
1 1 1 1
Courtney D. DiNardo, MD ; Gautam Borthakur, MD ; Sherry A. Pierce, BA, BS, RN ; Elias Jabbour, MD ;
Sa A. Wang, MD2; Carlos Bueso-Ramos, MD, PhD2; Sanam Loghavi, MD2; Guillin Tang, MD, PhD2; Cora M. Cheung, RN1;
Lynette Alexander, BA, RN1; Steven Kornblau, MD1; Michael Andreeff, MD, PhD1; Guillermo Garcia-Manero, MD 1
;
Farhad Ravandi, MD1; Marina Y. Konopleva, MD, PhD1; and Naval Daver, MD 1

BACKGROUND: Patients with relapsed/refractory (R/R) acute myeloid leukemia (AML) have limited treatment options. In preclinical
models of AML, inhibition of the PD-1/PD-L1 axis demonstrated antileukemic activity. Avelumab is an anti–PD-L1 immune checkpoint
inhibitor (ICI) approved in multiple solid tumors. The authors conducted a phase 1b/2 clinical trial to assess the safety and efficacy of
azacitidine with avelumab in patients with R/R AML. METHODS: Patients aged ≥18 years who had R/R AML received azacitidine 75 mg/
m2 on days 1 through 7 and avelumab on days 1 and 14 of 28-day cycles. RESULTS: Nineteen patients were treated. The median age was
66 years (range, 22-83 years), 100% had European LeukemiaNet 2017 adverse-risk disease, and 63% had prior exposure to a hypometh-
ylating agent. Avelumab was dosed at 3 mg/kg for the first 7 patients and at 10 mg/kg for the subsequent 12 patients. The most common
grade ≥3 treatment-related adverse events were neutropenia and anemia in 2 patients each. Two patients experienced immune-related
adverse events of grade 2 and grade 3 pneumonitis, respectively. The overall complete remission rate was 10.5%, and both were complete
remission with residual thrombocytopenia. The median overall survival was 4.8 months. Bone marrow blasts were analyzed for immune-
related markers by mass cytometry and demonstrated significantly higher expression of PD-L2 compared with PD-L1 both pretherapy
and at all time points during therapy, with increasing PD-L2 expression on therapy. CONCLUSIONS: Although the combination of azaciti-
dine and avelumab was well tolerated, clinical activity was limited. High expression of PD-L2 on bone marrow blasts may be an important
mechanism of resistance to anti–PD-L1 therapy in AML. Cancer 2021;127:3761-3771. © 2021 American Cancer Society.

LAY SUMMARY:
• This report describes the results of a phase 1b/2 study of azacitidine with the anti–PD-L1 immune checkpoint inhibitor avelumab for
patients with relapsed/refractory acute myeloid leukemia (AML).
• The clinical activity of the combination therapy was modest, with an overall response rate of 10.5%.
• However, mass cytometry analysis revealed significantly higher expression of PD-L2 compared with PD-L1 on AML blasts from all
patients who were analyzed at all time points.
• These data suggest a novel potential role for PD-L2 as a means of AML immune escape.

KEYWORDS: avelumab, azacitidine, checkpoint inhibitor, mass cytometry, PD-1, PD-L1, PD-L2.

INTRODUCTION
Over the past 4 years, multiple new therapies have been US Food and Drug Administration-approved for the treatment
of acute myeloid leukemia (AML).1 Despite these advancements, outcomes for the majority of patients with AML who
do not undergo allogeneic stem cell transplantation (allo-SCT) remain dismal, especially for those patients with relapsed/
refractory (R/R) disease who have an expected median overall survival (OS) of 4 to 7 months.2-4
Since the initial approval of the anti–CTLA-4 antibody ipilimumab in 2011 for melanoma, multiple immune
checkpoint inhibitors (ICIs) targeting CTLA-4, PD-1, and PD-L1 have dramatically improved outcomes and
have been approved in many solid tumors.5 Efficacy of ICIs for hematologic malignancies has generally been less
impressive, with US Food and Drug Administration approvals limited thus far to Hodgkin lymphoma and primary
mediastinal B-cell lymphoma.6 No ICI has received approval for leukemia, and trials assessing ICI-based therapy
for AML have only recently been presented and published, with acceptable safety profiles but generally modest

Corresponding Author: Naval Daver, MD, Department of Leukemia, The University of Texas MD Anderson Cancer Center, 1515 Holcombe Boulevard, Unit 0428, Houston,
TX 77030 ([email protected]).
1
Department of Leukemia, The University of Texas MD Anderson Cancer Center, Houston, Texas; 2 Department of Hematopathology, The University of Texas MD Anderson
Cancer Center, Houston, Texas
The first 2 authors contributed equally to this article.
Additional supporting information may be found in the online version of this article.

DOI: 10.1002/cncr.33690, Received: February 1, 2021; Revised: April 17, 2021; Accepted: May 3, 2021, Published online June 25, 2021 in Wiley Online Library
(wileyonlinelibrary.com)

Cancer October 15, 2021

53
Original Article

efficacy.7-12 Several reasons have been postulated for the may lead to different efficacy profiles.5,33 The current
limited efficacy of ICIs in AML compared with that phase 1b/2 study was designed to assess the combina-
in solid tumors and Hodgkin lymphoma. The protec- tion of azacitidine with avelumab, an anti–PD-L1 ICI,
tive bone marrow (BM) microenvironment might exert in patients with R/R AML.
an immunosuppressive influence by preventing access
of T cells to AML blasts or by secretion of immune- MATERIALS AND METHODS
dampening metabolites such as IDO or arginine.13-16
Patient Eligibility
ICI efficacy typically correlates with the tumor muta-
Patients who had AML that was refractory (up to
tional burden.17 The tumor mutational burden of AML
3 prior therapies for AML) or relapsed (up to salvage
is logarithmically lower compared with that of many
3 status) who were aged ≤18 years with an Eastern
solid tumor malignancies.18 However, a critical role for
T-cell activity in the control of AML has been leveraged Cooperative Oncology Group performance status ≤2
for over 40 years since early demonstrations of a graft- and adequate organ function, defined as total biliru-
versus-leukemia effect with allo-SCT and subsequently bin ≤1.5 × the upper limit of normal (ULN) (≤3 ×
with donor lymphocyte infusions post-SCT.19-21 This ULN if considered caused by leukemic involvement or
suggests that a broad and comprehensive evaluation of Gilbert syndrome), aspartate aminotransferase and ala-
immune strategies is warranted before final conclusions nine aminotransferase levels ≤2.5 × ULN (≤5 × ULN
regarding the efficacy and applicability of various im- if considered caused by leukemic involvement), and an
mune modalities in AML can be drawn. estimated creatinine clearance >30 mL per minute (as
Immune evasion in AML is likely a multifaceted calculated by the Cockcroft-Gault formula or similar
process.22 ICI therapy for AML was first reported as institutional standard method), were eligible. Prior
part of a phase 1/1b clinical trial, with striking anti- therapy for myelodysplastic syndrome (MDS), chronic
leukemic efficacy observed (especially for extramed- myelomonocytic leukemia, or a myeloproliferative neo-
ullary disease) in a subset of patients with AML who plasm was not considered as prior AML therapy. Key
had post–allo-SCT relapse treated with high-dose ip- exclusion criteria included known severe hypersensi-
ilimumab.8 Human AML cells express various amounts tivity to monoclonal antibodies, uncontrolled asthma,
of transcripts for the PD-1 ligands PD-L1 and PD-L2, known history of severe interstitial lung disease or
and higher messenger RNA (mRNA) expression of pneumonitis, prior exposure to another PD-1/PD-L1
PD-L1 and PD-L2 is correlated with inferior OS in pa- inhibitor in combination with azacitidine, any active
tients with AML.23-26 In a murine model of AML, both autoimmune disease that could deteriorate with treat-
PD-1 gene knockout as well as anti–PD-L1 murine ment, and prior organ allograft (with the exception of
ICI led to decreased leukemic burden and improved prior allo-SCT >3 months from initiation of protocol
survival.27-29 In patients with AML, BM-infiltrating therapy). The study was conducted in accordance with
T-cell populations appeared to be preserved compared the Declaration of Helsinki, and all participants signed
with BMs from healthy individuals, with an increased a written informed consent document (ClinicalTrials.
frequency of immune inhibitory and activating core- gov identifier NCT02953561). The complete protocol
ceptors (especially in relapsed AML), including PD-1, is attached as Supporting Materials.
OX40, and TIM3.30,31 Further treatment with the com-
monly used hypomethylating agent (HMA) azacitidine Study Design and Objectives
increased expression of PD-1, PD-L1, and PD-L2 in pa- This was a phase 1b/2, nonrandomized, single-center,
tients with myeloid malignancies.25,32 On the basis of open-label study evaluating the safety and efficacy of
these data, a phase 2 trial evaluated the efficacy of azac- avelumab in combination with azacitidine for pa-
itidine with nivolumab, an anti–PD-1 ICI, in patients tients with R/R AML. Patients were recruited between
with R/R AML.7 In that study of 70 patients who had February 2017 and May 2018. The data cutoff date was
R/R AML, the overall response rate (ORR) was 33%. June 1, 2020. The primary objective of the phase 1b
The median OS was especially encouraging in salvage 1 portion was to determine the maximum tolerated dose
patients (10.5 months), which was superior to matched (MTD) and dose-limiting toxicity (DLT) of the com-
historical controls. Although PD-1 and PD-L1 partici- bination. Definitions for DLTs and immune-related
pate in the same axis, blockade of the tumor cell ligand adverse events (irAEs) are outlined in the clinical pro-
(PD-L1) rather than the effector T-cell receptor (PD-1) tocol in sections 5.2.1.3 and 5.3.5, respectively (see

Cancer October 15, 2021

54
Azacitidine and Avelumab in AML/Saxena et al

Supporting Materials). The primary objective of the to the Common Terminology Criteria for Adverse Events,
phase 2 portion of the study was to define the ORR, version 4.03.
defined as complete remission (CR) + CR with incom-
plete platelet recovery (CRp) + CR with incomplete Statistical Methods
blood count recovery (CRi) + morphologic leukemia- Futility and toxicity were assessed using the Bayesian ap-
free state according to the AML International Working proach of Thall and Sung.35 Patient demographics were
Group (IWG) 2003 response criteria.34 Secondary ob- analyzed using descriptive statistics, and survival analyses
jectives included the number of patients who achieved were performed using Kaplan-Meier methodology. EFS
>50% reduction in BM blast percentage while on ther- was calculated as time from treatment initiation to change
apy, event-free survival (EFS), and OS. Exploratory ob- in AML therapy, patient death from any cause, or loss
jectives included evaluation of minimal residual disease to follow-up. OS was calculated as the time from treat-
by multiparametric flow cytometry and longitudinal ment initiation to death from any cause. For comparisons
analysis of immunological markers on peripheral blood of PD-L1 and PD-L2 blast expression, statistical analy-
(PB) and BM aspirate AML blasts. AML blasts were ses were performed using the Wilcoxon matched-pairs
assessed for expression of multiple markers (including signed-rank test. For comparisons of OS based on salvage
PD-L1, PD-L2, PD-1, CTLA-4, TIGIT, OX40, TIM3, status and prior HMA exposure, statistical analyses were
CD200, LAG3, 4-1BB) with a mass cytometry/cytom- performed using the Mann-Whitney test.
etry by time-of-flight (CyTOF) panel of antibodies de-
signed and conjugated for this study (see Supporting Immunophenotyping of BM and PB Samples by
Mass Cytometry
Methods), as described below.
Frozen primary BM samples were thawed and immedi-
Treatment Regimen and Safety Assessment ately incubated in thawing media (10 mL Eagle’s mini-
Azacitidine was administered on days 1 through 7 intrave- mum essential medium, α modification; Sigma Life
nously or subcutaneously at a dose of 75 mg/m2, and ave- Science) with 20% heat-inactivated fetal bovine serum
lumab was administered on days 1 and 14 intravenously (GenDEPOT), 10 mM MgSO4, 100 μg/mL heparin,
of each 28-day cycle. In the phase 1b dose-escalation and DNAse for 15 minutes at 37 °C before CyTOF stain-
portion, cohorts of 6 patients were enrolled in progres- ing. Sample barcoding and metal-conjugated antibody
sively increasing doses of avelumab with standard dose staining were performed according to Fluidigm protocols
azacitidine to identify the MTD and the recommended (for additional details, including the customized panel of
phase 2 dose (RP2D) for the phase 2 portion. Avelumab antibodies used, see Supporting Methods).
was initially administered at a −1 dose of 3 mg/kg with
Mass Cytometry Data Analysis
standard-dose azacitidine to the first 7 patients enrolled,
Data were first demultiplexed using Fluidigm Debarcoder
of whom 6 were evaluable for DLTs for 28 days, as speci-
software. Individual mass cytometry data files (.fcs) were
fied in the protocol. Because no DLTs were identified, the
then filtered using FlowJo to remove the normalization
subsequent 6 patients received avelumab 10 mg/kg with
beads, debris, doublets, and dead cells. Remaining analy-
standard-dose azacitidine, again with no DLTs, and this
ses were performed in R (version 3.6.1; The R Foundation
was identified as the MTD and selected as the RP2D for
for Statistical Computing) using the R packages cytofkit36
the combination. Up to 40 additional patients (excluding
and flowcore.37 Processed data were subjected to negative
patients treated at the RP2D from the lead-in part) could
value-pruned, inverse hyperbolic sine transformation and
be recruited for the phase 2 part. However, only an ad-
clustered based on the PhenoGraph algorithm (k = 22)
ditional 6 patients were enrolled in the expansion phase 2
using all cell surface markers.38 Dimensionality reduction
at the RP2D, for a total of 12 patients (6 from the DLT
was performed using the uniform manifold approxima-
evaluation phase 1b and 6 in the expansion phase 2) who
tion and projection method.39
received azacitidine with avelumab 10 mg/kg. The study
was terminated early because of modest efficacy and other
competing protocol priorities within the institution. Dose RESULTS
interruptions of both azacitidine and avelumab were per- Patient Characteristics and Treatment
mitted, as were dose reductions or modifications of azac- Between February 2017 and May 2018, 19 patients with
itidine (for details of the full protocol, see the Supporting R/R AML were enrolled and treated. These included
Materials). Adverse events (AEs) were defined according 6 patients in the safety cohort. Patient characteristics are

Cancer October 15, 2021

55
Original Article

TABLE 1. Baseline Patient Characteristics at the TABLE 2. Treatment-Related Adverse Events

Time of Treatment Initiation, N = 19 Occurring in ≥10% of Patients

No. of Patients (%) No. of Patients (%)

Characteristic or Median [Range]
Adverse Event Any Grade Grade ≥3
Age, y 66 [22-83]
≥60 y 14 (74) Diarrhea 5 (26.3) 1 (5.3)
Sex Fatigue 5 (26.3) 1 (5.3)
Women 9 (47) Nausea 5 (26.3) 1 (5.3)
Diagnosis Infusion-related reaction 4 (21.1)
De novo AML 6 (32) Oral mucositis 4 (21.1)
Secondary AML 13 (68) Anorexia 3 (15.8) 1 (5.3)
Prior HMA exposure for MDS or AML 12 (63) Constipation 3 (15.8)
No. of prior regimens for AMLa 1 [1-3] Anemia 3 (15.8) 2 (10.5)
Prior regimens for AML Increased ALT 2 (10.5)
HMA-based 6 (32) Chills 2 (10.5)
HiDAC 8 (42) Hypertension 2 (10.5)
IDAC 10 (53) Maculopapular rash 2 (10.5)
Targeted therapiesb 7 (37) Muscle weakness 2 (10.5)
Prior allogeneic SCT 3 (16) Pleural effusion 2 (10.5)
BM blasts, % 40 [7-88] Pneumonitis 2 (10.5) 1 (5.3)
White blood cell count, ×109/L 2.4 [0.7-39.4] Leukopenia 2 (10.5)
Platelets, ×109/L 22 [2-67] Neutropenia 2 (10.5) 2 (10.5)a
ELN risk classification: Cytogenetics and Lymphopenia 2 (10.5) 1 (5.3)
mutations
Favorable 0 (0) Abbreviation: ALT, alanine aminotransferase.
a
Intermediate 0 (0) One of these was grade 4 neutropenia and was the only grade 4 treatment-
Adverse 19 (100) related adverse event. No grade 5 treatment-related adverse events were
Cytogenetics by ELN classification documented.
Favorable 0 (0)
Intermediate 5 (26)
Adverse 14 (74)
Molecular mutational panelc HMA exposure, 6 had prior azacitidine exposure, and 6
TP53 7 (37)
TET2 5 (26) had prior HMA exposure for AML. Three patients had
ASXL1 5 (26) prior allo-SCT, including 2 who had 2 prior allo-SCTs.
CEBPA 2 (11)
RAS 4 (21) TP53 was the most common mutation (37%) identified.
GATA2 2 (11) All patients who received at least 1 dose of azaciti-
SRSF2 3 (16)
SF3B1 2 (11)
dine and 1 dose of avelumab were evaluable for efficacy
RUNX1 5 (26) and safety per an intention-to-treat approach.
Abbreviations: AML, acute myeloid leukemia; BM, bone marrow; ELN,
European LeukemiaNet; HiDAC, high-dose cytarabine based; HMA, hypo- Safety
methylating agent; IDAC, intermediate-dose cytarabine-based; MDS, myelo-
All treatment-related AEs (TRAEs) occurring in
dysplastic syndrome; SCT, stem cell transplantation.
a
For patients who had relapsed acute myeloid leukemia after allogeneic SCT, ≥10% of patients (ie, at least 2 patients) are listed in
the included regimens were received before allogeneic SCT.
b
Table 2. The most common TRAEs were diarrhea (n = 5),
Targeted therapies included: IDH1 inhibitor, anti-CD33 antibody-drug conju-
gate, BET inhibitor, SYK inhibitor, Grb-2 antisense oligonucleotide, and SMAC fatigue (n = 5), nausea (n = 5), infusion reactions (n = 4),
mimetic.
c
and mucositis (n = 4). IrAEs occurred in only 2 patients
The mutations listed were present in ≥2 patients.
(10.5%), and both were pneumonitis (1 grade 3 and 1
grade 2). An additional patient developed grade 2 pneu-
summarized in Table 1. The median age at enrollment monia/pneumonitis attributed to cytomegalovirus, which
was 66 years (range, 22-83 years), and 68% of patients was confirmed by viral culture from a bronchoalveolar
had secondary AML. This was a high-risk population: lavage sample, and she was treated with valganciclovir.
all 19 patients had adverse-risk disease according to the Both patients with possible ICI-related pneumonitis were
ELN 2017 risk stratification criteria, and 74% of patients diagnosed clinically based on hypoxemia in the presence
had adverse cytogenetics according to ELN 2017 crite- of radiographic findings on computed tomography imag-
ria.40 The median number of prior treatment regimens for ing and no evidence of an alternative explanation after an
AML was 1 (range, 1-3 prior treatments). Twelve (63%) extensive workup (such as cardiac, pulmonary, or infec-
patients had prior HMA exposure with azacitidine, decit- tious). The first patient with an irAE experienced grade 3
abine, or SGI-110 (an investigational prodrug of decit- pneumonitis during cycle 3, 98 days after study entrance
abine) for MDS or AML. Of the 12 patients with prior (she received both doses of avelumab during cycle 1 and

Cancer October 15, 2021

56
Azacitidine and Avelumab in AML/Saxena et al

no more doses of avelumab after cycle 1 because of an un- per patient was 5 (range, 1-14 doses). Reasons for study
derlying fungal pneumonia, which immediately preceded discontinuation were lack or loss of response with sub-
cycle 1 of azacitidine plus avelumab). During cycle 2, the sequent therapy change (n = 8), lack or loss of response
patient developed a bacterial pneumonia. During cycle 3, without subsequent therapy change (n = 3), transition to
the patient developed worsening bilateral lung ground- hospice care (n = 3), therapy change to ease travel needs
glass opacities and hypoxemia without a clear infectious or per patient preference (n = 1), and death while on treat-
cardiac etiology despite not having received avelumab for ment (n = 4).
84 days. In the absence of a clear alternative etiology, this
event was attributed as a possible irAE from avelumab. Responses and Survival
She received steroids for 36 days, was hospitalized for 9 The ORR according to IWG 2003 criteria was 10.5%,
days after pneumonitis onset, and was taken off study with 2 patients achieving CRp, as indicated in Table 3.
after this event because of lack of response to azacitidine Both patients were positive for minimal residual disease
plus avelumab. The second patient experienced grade 2 by multiparametric flow cytometry through the course
pneumonitis on day 9 during cycle 1 (after the first dose of their response. The first patient had secondary AML,
of avelumab). He received steroids for 7 days, was hospi- diploid cytogenetics, and mutations in TET2 and RUNX1
talized for 16 days after pneumonitis onset, and remained and was initially treated with cladribine plus SGI-110
on study with azacitidine alone without further avelumab (a prodrug of decitabine) without sufficient response,
exposure. An additional patient developed grade 3 diar- and he remained transfusion-dependent on platelets. He
rhea of unclear etiology that self-resolved within 3 days received azacitidine plus avelumab (3 mg/kg cohort) with
without steroid or immunosuppressive therapy and thus 7% BM blasts pretreatment. He achieved CRp with 2%
was not considered an irAE but was attributed as possibly blasts at the end of cycle 1 and had 0% blasts at the end of
(Common Terminology Criteria for Adverse Events attri- cycle 3. He continued on therapy for 6 cycles (7.6 months
bution is a score of 3 for possible) related to azacitidine. on treatment), remaining red blood cell transfusion-
A separate patient developed grade 3 colitis, which was independent but platelet transfusion-dependent. During
attributed to Clostridium difficile infection based on posi- cycle 5, the patient was beginning to lose response, with
tive stool C. diff DNA testing. There was 1 grade 4 TRAE the emergence of 1% to 2% blasts in the PM (BM was
(neutropenia), and no grade 5 TRAEs were reported. All deferred to confirm relapse). After cycle 6, he became newly
AEs of any grade and frequency, regardless of attribution, transfusion-dependent on red blood cells in addition to
are listed in Supporting Table 1 and all serious AEs are platelets. Treatment was discontinued without subsequent
listed in Supporting Table 2. The most common AEs, therapy change because of patient preference to transition
irrespective of attribution, were constipation (n = 12), to supportive care locally (patient was from out of state).
fatigue (n = 11), and muscle weakness (n = 11). The The second patient with an objective response had de novo
most common grade ≥3 AEs, irrespective of attribution, AML, complex cytogenetics, and mutations in ASXL1,
were neutropenic fever (n = 6) and pneumonia (n = 5). SRSF2, SETBP1, and RUNX1. He was refractory to 7 + 3
The median number of cycles of therapy received and started azacitidine plus avelumab (10 mg/kg) with 35%
was 3 (range, 1-7 cycles), and the median time on study BM blasts pretreatment. The patient achieved CRp with
was 3.3 months. The median duration of cycle 1 and 2 4% blasts at the end of cycle 1 and had 4% blasts at the end
was 28 and 30 days, respectively. Azacitidine was held for of cycle 2. The patient experienced grade 2 pneumonitis
at least 1 dose in 5 patients for the following indications: after his first dose of avelumab and did not receive further
neutropenic fever (n = 3), patient preference to stop ther- avelumab for the remainder of the study. He did not have
apy and transition to hospice care (n = 1), and travel dif- further BM biopsies and continued on study protocol with
ficulties (n = 1). Avelumab was held for at least 1 dose in azacitidine alone for a total of 5 cycles (5 months). The
13 patients for the following indications (some patients patient came off protocol because of difficulty traveling
had doses held for multiple indications): pneumonia and continued azacitidine monotherapy closer to home for
(n = 2), patient preference (n = 2), travel/scheduling dif- at least 4 more months. He died from disease progression
ficulties (n = 2), pneumonitis (n = 1), infusion reaction approximately 6 months after coming off protocol.
(n = 1), neutropenic fever (n = 1), elevated creatinine Three additional patients had BM blast reductions of
(n = 1), hyperbilirubinemia (n = 1), hypercalcemia >50% from pretherapy (51% → 25%, 32% → 12%, and
(n = 1), and deconditioning (n = 1) (see Supporting 40% → 13%; median time to >50% blast reduction, 26
Table 3). The median number of avelumab doses received days) that did not meet IWG criteria for a partial response

Cancer October 15, 2021

57
Original Article

TABLE 3. Best International Working Group A Event-free survival

Response Attained on Trial, N = 19
100
No. of N Median EFS

Event-free survival (%)

Characteristic Patients(%)
80 19 3.6 months
Overall response rate 2 (10.5)
CR 0 (0.0)
CRi/CRp 2 (10.5)a 60
PR/MLFS 0 (0.0)
50% Blast reduction without count recoveryb 3 (15.8) 40
Stable disease >6 moc 1 (5.3)
Nonresponders 13 (68.4)
Median no. of cycles to best response 20
CRi/CRp 1
Blast reduction: Median [range]b 1 [1-3] 0
30-Day mortality 2 (10.5)
0 5 10 15
60-Day mortality 4 (21.0)
Months
Abbreviations: CR, complete remission; CRi, complete remission with incom-
plete count recovery; CRp, complete remission with residual thrombocytope-
nia; MLFS, morphologic leukemia-free state; PR, partial response. B Overall survival
a
Both patients had CRp and were positive for minimal residual disease by
flow cytometry through response. 100
b
Blast reduction without count recovery was defined as a reduction in the N Median OS
bone marrow blast percentage by at least one-half without recovery to an 19 4.8 months

Overall survival (%)

absolute neutrophil count ≥1000/μL and a platelet count ≥100,000/μL.
80
c
Stable disease was defined as the absence of CR, CRi, PR, MLFS, hema-
tologic improvement without evidence of clinical deterioration or progressive 60
disease, maintained >6 months on study. The stable disease in this patient
was maintained for 7.8 months.
40

20
or for CR/CRi/CRp, and 1 patient had stable disease for
7.8 months. The remaining 13 patients had no evidence 0
of response or clinical benefit. The 30-day mortality was 0 5 10 15
11%; 1 patient died on day 4 from rapidly progressive dis- Months
ease, and one patient died on day 21 from sepsis secondary
to pneumonia. An additional 2 deaths occurred between Figure 1. Survival analyses, including (A) event-free survival
(EFS) and (B) overall survival (OS), are illustrated for all 19
treatment days 31 and 60; 1 from rapidly progressive dis- patients who were treated on the clinical trial of azacitidine
ease after 1 cycle with transition to hospice care and the plus avelumab.
other during cycle 2 from unknown causes at home. At
the time of data analysis, all 19 patients had died. The
median EFS was 3.6 months (range, 0.1-10.2 months) assessed, there were significantly more BM blasts ex-
(Fig. 1A). The median OS was 4.8 months (range, 0.1- pressing PD-L2 than PD-L1 (median, 34.9% vs 12%;
11 months) (Fig. 1B). The median OS in salvage 1 versus P < .005) (Table 4). Double-positive (PD-L1-positive/
salvage >1 patients was 5.8 versus 4.6 months (P = .84), PD-L2-positive) blasts accounted for only 4% (median)
respectively. The median OS in TP53-mutated patients of BM blasts. Similar to BM blasts, PB blasts displayed
was 4.8 months. The median OS in patients with prior higher expression of PD-L2 versus PD-L1 (Figs. 2 and 3).
HMA exposure versus HMA-naive patients was 5.7 versus Next, we assessed whether the percentage of BM blasts
4.6 months (P = .53), respectively. The 2 patients who expressing PD-L2 and PD-L1 changed during treatment
achieved CRp had an OS of 10.2 and 11 months. with azacitidine plus avelumab by performing mass cytom-
etry on serially obtained BM aspirate samples from 8 pa-
Immune Profiling by Mass Cytometry tients. For all 8 patients, PD-L2 expression remained more
To characterize cellular markers that may predict ICI ef- abundant than PD-L1 expression at each time point (see
ficacy, we performed mass cytometry (CyTOF) on BM Supporting Fig. 1). Finally, we evaluated BM blast expres-
aspirate and PB samples from 9 patients who were treated sion of other immune-related markers, including OX40,
on study. We first assessed pretherapy PD-L1 and PD- LAG3, PD-1, and CTLA-4. Expression of these markers was
L2 surface expression by performing CyTOF on nonper- assessed pretherapy and compared with expression on BM
meabilized BM and PB blasts (Fig. 2). For all patients blasts from samples obtained before trial discontinuation

Cancer October 15, 2021

58
Azacitidine and Avelumab in AML/Saxena et al

Figure 2. Pretreatment PD-L1 and PD-L2 expression levels are illustrated on peripheral blood (PB) blasts and bone marrow (BM)
blasts in 6 patients. Displayed are the percentages of acute myeloid leukemia blasts that were PD-L1–positive versus PD-L2–positive
in (Left) BM and (Right) PB in pretreatment samples from 6 different patients.

TABLE 4. Immunophenotype of Bone Marrow Blasts Pretreatment With Azacitidine Plus Avelumab

% PD-L1+/PD-L2+ Best Response

Patient % Bone Marrow Blasts % PD-L1+ % PD-L2+ (Double Positive) on Therapy
1 28 12.0 50.1 6.6 Stable disease
2 53 16.2 64.3 10.7 No response
3 27 9.0 34.9 3.2 No response
4 59 8.2 26.9 2.4 No response
5 7 7.1 26.3 4.0 CRp
6 51 3.1 16.9 0.6 Blast reduction
7 45 12.4 32.9 4.0 No response
8 38 21.9 53.3 11.3 No response
9 57 31.6 62.4 21.3 No response

Abbreviations: +, positive; CRp, complete remission with residual thrombocytopenia.

to assess increase, decrease, or stable expression of the im- lymphocyte infusion, as well as more recent data dem-
mune marker on BM blasts (see Supporting Fig. 2). We onstrating early efficacy signals of CTLA-4 and PD-1
noted that PD-L2 was the most frequent immune marker ICIs in AML, we conducted a study to assess the com-
to be increased by >5% from baseline on BM blasts (in bination of azacitidine and the anti–PD-L1 antibody
5 of 8 patients) during treatment. No other correlations avelumab in patients with R/R AML. The regimen was
were found in the change in blast surface immune-related well tolerated. Four patients died while on active treat-
marker expression during treatment. Most of the immune ment; no deaths were directly attributable to treatment.
checkpoints interrogated were present on only a subset of Only 2 patients had irAEs (both pneumonitis), and 1
BM blasts in each patient, and PD-L2 (median, 34%), patient permanently discontinued avelumab treatment
TIGIT (median, 20.5%), and CTLA-4 (median, 18.5%) because of the irAE (grade 2 pneumonitis). Notably,
were the most abundant checkpoints expressed at baseline. this patient was 1 of only 2 patients who experienced
Given that only 2 of the patients analyzed had evidence an objective response, with a BM blast reduction from
of an antileukemic response to treatment (2 of 8 evaluated 35% to 4%. Although the patient only received 1 dose
[patients 5 and 6]), we did not perform analysis comparing of avelumab (after which he developed pneumonitis),
responders with nonresponders. he maintained a clinical response to azacitidine mono-
therapy for several more cycles, suggesting the possibil-
DISCUSSION ity of immune-mediated disease control. In solid tumor
On the basis of historical evidence of a T-cell– malignancies, the occurrence of irAEs may correlate
mediated antileukemic effect from allo-SCT and donor with enhanced response to ICI therapy.41

Cancer October 15, 2021

59
Original Article

Figure 3. Representative PD-L1/PD-L2 baseline expression in bone marrow (BM) and peripheral blood (PB) are shown. This
representative cytometry by time-of-flight plot from a patient’s BM and PB illustrates relative PD-L1 and PD-L2 expression on blasts.

The ORR in this study was 11% (both CRp), Interestingly, these results do not appear unique to
with a median OS of 4.8 months in all patients and a nivolumab or avelumab. Two additional phase 2 trials
median OS of 5.7 months in previously HMA-naive of ICI for AML recently performed at other centers also
patients. This efficacy is comparable to the historical suggest contrasting efficacy between anti–PD-1 and anti–
CR/CRi rate of 16% and the median OS of 6.7 months PD-L1 therapy. In a phase 2 study of the anti–PD-1 ICI
achieved with azacitidine monotherapy in a large co- pembrolizumab with azacitidine (ClinicalTrials.gov iden-
hort of previously HMA-naive patients with R/R AML tifier NCT02845297), 2 cohorts were enrolled: 1 cohort
and suggests that the addition of avelumab did not of patients with R/R AML and 1 cohort of treatment-naive
add clinical benefit.4 These results stand in contrast to patients who were ineligible for intensive chemotherapy.9
those from a previous phase 2 study of azacitidine plus In the R/R AML cohort, the ORR was 32% (CR/CRi,
nivolumab for R/R AML (ClinicalTrials.gov identifier 14%) with median OS of 10.8 months, similar to results
NCT02397720) in which the ORR was 33% (CR/ from the azacitidine plus nivolumab study.9 Notably, the
CRi rate, 22%), the median OS in all patients was 6.3 treatment-naive cohort had an ORR of 71% (CR/CRi,
months, and the median OS in salvage 1 patients was 47%) and a median OS of 13.1 months.9 A separate inter-
10.6 months.7 In that study, previously HMA-naive pa- national, randomized phase 2 study compared azacitidine
tients had a very encouraging ORR of 52%. Both the with or without the anti–PD-L1 durvalumab for front-
current study and the azacitidine plus nivolumab study line treatment-naive patients who had higher risk MDS
were conducted in a similar patient population during or AML (ClinicalTrials.gov identifier NCT02775903).11
similar time frames at the same institution: median That study included 129 patients with AML: 64 patients
age, 66 versus 70 years; median 1 versus 2 prior lines of received azacitidine with durvalumab, and 65 received
therapy; 63% versus 64% of patients with prior HMA azacitidine alone.11 The CR/CRi rates were similar in
exposure; median pretreatment BM blast percentage, both arms (31% vs 35%, respectively), as was the median
35% versus 40%; and 16% versus 19% with prior allo- OS (13 vs 14.4 months, respectively), suggesting no ben-
SCT.7 Thus the main apparent difference between these efit from the addition of durvalumab.11 Thus, based on
2 studies, with the caveats of cross-trial comparisons, these 4 publicly presented studies of azacitidine with ICI
appears to be use of the anti–PD-1 nivolumab in the for patients with AML, 2 conclusions may be suggested.
prior trial and use of the anti–PD-L1 avelumab in the First, azacitidine with an ICI (anti–PD-1 or anti–PD-L1),
current trial. as with almost all other AML therapies, appears to be

Cancer October 15, 2021

60
Azacitidine and Avelumab in AML/Saxena et al

more effective in treatment-naive patients than in patients expression, 12%). Given that these patients had not had
with R/R disease. Second, regardless of frontline or R/R prior ICI exposure pretreatment, these data suggest that
disease, anti–PD-L1 therapy appears to generate inferior the elevated baseline PD-L2:PD-L1 ratio was not due to
response and OS compared with anti–PD-1 therapy in selective pressure by treatment with an anti–PD-L1 anti-
combination with azacitidine. It is important to note that body. Although surface expression patterns varied during
both of these conclusions must be interpreted within the treatment, all patients assessed had more blasts express-
limitations and constraints of cross-trial comparisons, and ing PD-L2 compared with PD-L1 at each time point.
with only 1 of the 4 studies including a randomized co- Furthermore, in nearly all patients at each time point
hort (azacitidine with or without durvalumab). examined, PD-L2 was the most frequently expressed of
Given that ICI therapy with PD-1 versus PD-L1 10 examined immune markers on BM blasts. PD-L2 was
blockade has not been directly assessed in a randomized also the most common immune marker for which BM
trial for any malignancy, it must be acknowledged that expression increased by >5% on BM blasts from pretreat-
comparing these 2 approaches implements an inherently ment to end of treatment. Although our study is limited
biased method of comparing outcomes between clinical by a small sample size, and only a subset of patients had
trials. However, there is precedent for anti–PD-1 therapy BM aspirate samples with sufficient quality/quantity of
to lead to improved outcomes compared with anti–PD-L1 cells to undergo immune marker profiling by CyTOF, the
therapy. In a recently published meta-analysis of 19 ran- conserved findings across all examined patients at all time
domized controlled trials of ICI therapy with anti–PD-1 points of higher PD-L2 surface protein expression on BM
versus anti–PD-L1 for solid tumors, anti–PD-1–based blasts compared with PD-L1 appears notable. Together,
therapy resulted in improved OS and progression-free these data suggest that high pretherapy PD-L2 expression
survival compared with anti–PD-L1.42 It is possible and increasing on-therapy PD-L2 expression may have
that a similar phenomenon exists in AML. The basis for promoted PD-L2–mediated escape from anti–PD-L1
such a difference may be caused in part by the inability therapy. To our knowledge, this study provides the first
of anti–PD-L1 ICIs to block PD-L2. In the classic PD- characterization of PD-L2 surface expression on AML
1–mediated immune axis, tumor cells expressing PD-L1 blasts and provides a baseline observation on which to
and/or PD-L2 can engage the PD-1 receptor on circu- pursue future studies investigating PD-L2 as a potential
lating/infiltrating T cells, thereby promoting peripheral means of immune escape in AML.
immune tolerance of tumor cells.5 PD-L1 is expressed In conclusion, blockade of the PD-1/PD-L axis with
on numerous cell types, whereas PD-L2 is expressed pri- anti–PD-1 antibodies may be a superior method of dis-
marily on hematopoietic cells, including myeloid leuko- rupting peripheral immune tolerance than anti–PD-L1
cytes.43-46 Emerging data have shown that PD-L2 may ICIs for AML because of the high expression of PD-L2
play an important role in solid tumor immune evasion found on AML blasts. Future ICI studies in myeloid ma-
and that PD-L2 has a higher affinity for PD-1 than does lignancies may be best served by focusing on anti PD-1–
PD-L1.47,48 Because anti–PD-1 antibodies block the in- based therapies.
teraction between PD-1/PD-L1 and PD-1/PD-L2, both
ligands are blocked with these ICIs, whereas anti–PD-L1
FUNDING SUPPORT
antibodies leave PD-L2 free to engage PD-1 and poten- This work was supported in part by The University of Texas MD Anderson
tially allow immune evasion/tolerance. Cancer Center Support Grant (CA016672), The University of Texas MD
Anderson Cancer Center Leukemia Specialized Programs of Research
AML blasts have been shown to express PD-L1, PD- Excellence grant (CA100632), the Charif Souki Cancer Research Fund, the
L2, and PD-1 at the mRNA level, and coexpression of Dick Clark Immunotherapy Fund, and generous philanthropic contribu-
tions to the MD Anderson Moon Shots Program.
different immune checkpoint transcripts correlates with
inferior outcomes.25,26 Expression of mRNA does not al-
ways correlate with protein expression (including for PD- CONFLICT OF INTEREST DISCLOSURES
L1 and PD-L2) and, to our knowledge, PD-L2 surface Tapan M. Kadia reports grants from Amgen, Ascentage, AstraZeneca,
Bristol-Myers Squibb, Celgene, and Incyte; grants and personal fees
protein expression has not been extensively characterized from AbbVie, Genentech, Jazz Pharmaceuticals, and Pfizer; and personal
on AML blasts.45 We found that 8 of 9 patients assessed fees from Novartis outside the submitted work. Courtney D. DiNardo
reports grants from Calithera; grants and personal fees from AbbVie,
by mass cytometry had >25% PD-L2–positive pretreat- Agios, Celgene, Daiichi Sankyo, ImmuneOnc, and Novartis; and per-
ment blasts (median PD-L2 protein expression on blasts, sonal fees Bayer, Jazz Pharmaceuticals, MedImmune, and Notable Labs
outside the submitted work. Elias Jabbour reports research grants and
34.9%). By comparison, only 1 of 9 patients had >25% consultancy from AbbVie, Adaptive Biotechnology, Amgen, Bristol-
PD-L1–positive pretreatment blasts (median PD-L1 Myers Squibb, Pfizer, and Takeda outside the submitted work. Farhad

Cancer October 15, 2021

61
Original Article

Ravandi reports honoraria from Daichii, and honoraria and service on 6. Vaddepally RK, Kharel P, Pandey R, Garje R, Chandra AB. Review
an advisory board from Astellas and Novartis. Marina Konopleva re- of indications of FDA-approved immune checkpoint inhibitors
ports grants from Ablynx, Agios, Ascentage, AstraZeneca, Calithera, per NCCN Guidelines with the level of evidence. Cancers (Basel).
Cellectis, Eli Lilly, Rafael Pharmaceutical, and Sanofi; grants and other 2020;12:738. doi:10.3390/cancers12030738
support from AbbVie, F. Hoffman La-Roche, Forty-Seven, Genentech, 7. Daver N, Garcia-Manero G, Basu S, et al. Efficacy, safety, and bio-
and Stemline Therapeutics; and other support from Amgen, Kisoji, markers of response to azacitidine and nivolumab in relapsed/refractory
Reata Pharmaceutical outside the submitted work; and has a patent acute myeloid leukemia: a nonrandomized, open-label, phase II study.
(US 7,795,305 B2: “CDDO-Compounds and Combination Therapie”) Cancer Discov. 2019;9:370-383. doi:10.1158/2159-8290.CD-18-0774
with royalties paid to Reata Pharm, a patent (“Combination Therapy 8. Davids MS, Kim HT, Bachireddy P, et al. Ipilimumab for patients with
With a Mutant IDH1 Inhibitor and a BCL-2”) licensed to Eli Lilly, relapse after allogeneic transplantation. N Engl J Med. 2016;375:143-
and a patent (62/993,166: Combination of a MCL-1 Inhibitor and 153. doi:10.1056/NEJMoa1601202
Midostaurin, Uses and Pharmaceutical Compositions Thereof ”) pending 9. Gojo I, Stuart RK, Webster J, Zeidner JF. Multi-center phase 2 study
to Novartis. Naval Daver reports research funding from AbbVie, Amgen, of pembrolizumab (Pembro) and azacitidine (AZA) in patients with re-
Astellas, Bristol-Myers Squibb, Daiichi Sankyo, FATE Therapeutics, lapsed/refractory acute myeloid leukemia (AML) and in newly diagnosed
Genentech, Gilead, Glycomimetics, Hanmi, ImmunoGen, Karyopharm, (≥65 Years) AML patients [abstract]. Blood. 2019;134(suppl 1):832.
Newave, Novimmune, Pfizer, Sevier, Trillium, and Trovagene; personal 10. Ravandi F, Assi R, Daver N, et al. Idarubicin, cytarabine, and nivolumab
fees from AbbVie, Agios, Astellas, Bristol-Myers Squibb, Celgene, in patients with newly diagnosed acute myeloid leukaemia or high-
Daiichi Sankyo, Genentech, Gilead, KITE, Novartis, Pfizer, Servier, risk myelodysplastic syndrome: a single-arm, phase 2 study. Lancet
STAR Therapeutics, Syndax, Trillium, and Trovagene; grants from the Haematol. 2019;6:e480-e488. doi:10.1016/S2352-3026(19)30114-0
SagerStrong Foundation; grants and other support from Affymetrix; 11. Zeidan AM, Cavenagh J, Voso MT, Silverman LR. Efficacy and safety
personal fees from Blueprint Medicines, Incyte, LFB Biotechnologies, of azacitidine (AZA) in combination with the anti-PD-L1 durvalumab
Pacylex Pharmaceuticals, and Roche Diagnostics; other support from (Durva) for the front-line treatment of older patients (pts) with acute
Cellectis, Daiichi Sankyo, Plexxikon, and Samus Therapeutics; personal myeloid leukemia (AML) who are unfit for intensive chemotherapy
fees and other support from Celgene, DAVA Oncology, MustangBio, (IC) and pts with higher-risk myelodysplastic syndromes (HR-MDS):
and Novartis; and personal fees, nonfinancial support, and other support results from a large, international, randomized phase 2 study [abstract].
from AbbVie and Stemline Therapeutics outside the submitted work. Blood. 2019;134(suppl 1):829.
The remaining authors made no disclosures. 12. Zeidner JF, Vincent BG, Esparza S, Ivanova A. Final clinical results of
a phase II study of high dose cytarabine followed by pembrolizumab in
relapsed/refractory AML [abstract]. Blood. 2019;134(suppl 1):831.
AUTHOR CONTRIBUTIONS 13. Mussai F, De Santo C, Abu-Dayyeh I, et al. Acute myeloid leukemia
Kapil Saxena: Designed the study, collected and analyzed the data, creates an arginase-dependent immunosuppressive microenvironment.
and wrote the article. Shelley Herbrich: Designed the study, collected Blood. 2013;122:749-758. doi:10.1182/blood-2013-01-480129
and analyzed the data, and wrote the article. Naveen Pemmaraju: 14. Mussai F, Egan S, Higginbotham-Jones J, et al. Arginine dependence of
Enrolled patients. Tapan M. Kadia: Enrolled patients. Courtney D. acute myeloid leukemia blast proliferation: a novel therapeutic target.
DiNardo: Enrolled patients. Gautam Borthakur: Enrolled patients. Blood. 2015;125:2386-2396. doi:10.1182/blood-2014-09-600643
Sherry Pierce: Collected and analyzed the data. Elias Jabbour: Enrolled 15. Muller-Thomas C, Heider M, Piontek G, et al. Prognostic value of
patients. Sa A. Wang: Performed the molecular and cytogenetic analysis. indoleamine 2,3 dioxygenase in patients with higher-risk myelodys-
Carlos Bueso-Ramos: Performed the molecular and cytogenetic analy- plastic syndromes treated with azacytidine. Br J Haematol. 2020;190:
sis. Sanam Loghavi: Performed the molecular and cytogenetic analysis. 361-370. doi:10.1111/bjh.16652
Guillin Tang: Performed the molecular and cytogenetic analysis. Cora 16. Lamble AJ, Lind EF. Targeting the immune microenvironment in
M. Cheung: Enrolled patients. Lynette Alexander: Enrolled patients. acute myeloid leukemia: a focus on T cell immunity. Front Oncol.
Steven Kornblau: Enrolled patients. Michael Andreeff: Enrolled patients. 2018;8:213. doi:10.3389/fonc.2018.00213
Guillermo Garcia-Manero: Enrolled patients. Farhad Ravandi: Enrolled 17. Yarchoan M, Hopkins A, Jaffee EM. Tumor mutational burden and
patients. Marina Konopleva: Designed the study, enrolled patients, col- response rate to PD-1 inhibition. N Engl J Med. 2017;377:2500-2501.
lected and analyzed the data, and wrote the article. Naval Daver: Designed doi:10.1056/NEJMc1713444
the study, enrolled patients, collected and analyzed the data, and wrote the 18. Chalmers ZR, Connelly CF, Fabrizio D, et al. Analysis of 100,000
article. All authors contributed to data collection, reviewed and approved human cancer genomes reveals the landscape of tumor mutational bur-
the article, and shared final responsibility for the decision to submit the den. Genome Med. 2017;9:34. doi:10.1186/s13073-017-0424-2
article for publication. 19. Horowitz MM, Gale RP, Sondel PM, et al. Graft-versus-leukemia reac-
tions after bone marrow transplantation. Blood. 1990;75:555-562.
20. Weiden PL, Flournoy N, Thomas ED, et al. Antileukemic effect of graft-
versus-host disease in human recipients of allogeneic-marrow grafts. N Engl
REFERENCES J Med. 1979;300:1068-1073. doi:10.1056/NEJM197905103001902
1. Short NJ, Konopleva M, Kadia TM, et al. Advances in the treatment of 21. Schmid C, Labopin M, Nagler A, et al. Donor lymphocyte infusion in
acute myeloid leukemia: new drugs and new challenges. Cancer Discov. the treatment of first hematological relapse after allogeneic stem-cell
2020;10:506-525. doi:10.1158/2159-8290.CD-19-1011 transplantation in adults with acute myeloid leukemia: a retrospective
2. Ravandi F, Ritchie EK, Sayar H, et al. Vosaroxin plus cytarabine ver- risk factors analysis and comparison with other strategies by the EBMT
sus placebo plus cytarabine in patients with first relapsed or refractory Acute Leukemia Working Party. J Clin Oncol. 2007;25:4938-4945.
acute myeloid leukaemia (VALOR): a randomised, controlled, double- doi:10.1200/JCO.2007.11.6053
blind, multinational, phase 3 study. Lancet Oncol. 2015;16:1025-1036. 22. Teague RM, Kline J. Immune evasion in acute myeloid leukemia: cur-
doi:10.1016/S1470-2045(15)00201-6 rent concepts and future directions. J Immunother Cancer. 2013;1:1.
3. Roboz GJ, Rosenblat T, Arellano M, et al. International random- doi:10.1186/2051-1426-1-13
ized phase III study of elacytarabine versus investigator choice in 23. Brodska B, Otevrelova P, Kuzelova K. Correlation of PD-L1 surface ex-
patients with relapsed/refractory acute myeloid leukemia. J Clin Oncol. pression on leukemia cells with the ratio of PD-L1 mRNA variants and
2014;32:1919-1926. doi:10.1200/JCO.2013.52.8562 with electrophoretic mobility. Cancer Immunol Res. 2016;4:815-819.
4. Stahl M, DeVeaux M, Montesinos P, et al. Hypomethylating agents in doi:10.1158/2326-6066.CIR-16-0063
relapsed and refractory AML: outcomes and their predictors in a large 24. Brodska B, Otevrelova P, Salek C, Fuchs O, Gasova Z, Kuzelova K.
international patient cohort. Blood Adv. 2018;2:923-932. doi:10.1182/ High PD-L1 expression predicts for worse outcome of leukemia pa-
bloodadvances.2018016121 tients with concomitant NPM1 and FLT3 mutations. Int J Mol Sci.
5. Wei SC, Duffy CR, Allison JP. Fundamental mechanisms of immune 2019;20:2823. doi:10.3390/ijms20112823
checkpoint blockade therapy. Cancer Discov. 2018;8:1069-1086. 25. Yang H, Bueso-Ramos C, DiNardo C, et al. Expression of PD-L1,
doi:10.1158/2159-8290.CD-18-0367 PD-L2, PD-1 and CTLA4 in myelodysplastic syndromes is enhanced

Cancer October 15, 2021

62
Azacitidine and Avelumab in AML/Saxena et al

by treatment with hypomethylating agents. Leukemia. 2014;28:1280- analysis pipeline. PLoS Comput Biol. 2016;12:e1005112. doi:10.1371/
1288. doi:10.1038/leu.2013.355 journal.pcbi.1005112
26. Chen C, Liang C, Wang S, et al. Expression patterns of immune 37. Hahne F, LeMeur N, Brinkman RR, et al. flowCore: a Bioconductor
checkpoints in acute myeloid leukemia. J Hematol Oncol. 2020;13:28. package for high throughput flow cytometry. BMC Bioinformatics.
doi:10.1186/s13045-020-00853-x 2009;10:106. doi:10.1186/1471-2105-10-106
27. Zhang L, Gajewski TF, Kline J. PD-1/PD-L1 interactions inhibit anti- 38. Levine JH, Simonds EF, Bendall SC, et al. Data-driven phenotypic dis-
tumor immune responses in a murine acute myeloid leukemia model. section of AML reveals progenitor-like cells that correlate with progno-
Blood. 2009;114:1545-1552. doi:10.1182/blood-2009-03-206672 sis. Cell. 2015;162:184-197. doi:10.1016/j.cell.2015.05.047
28. Zhou Q, Munger ME, Highfill SL, et al. Program death-1 signaling and 39. Becht E, McInnes L, Healy J, et al. Dimensionality reduction for vi-
regulatory T cells collaborate to resist the function of adoptively trans- sualizing single-cell data using UMAP. Nat Biotechnol. 2019;37:38-44.
ferred cytotoxic T lymphocytes in advanced acute myeloid leukemia. doi:10.1038/nbt.4314
Blood. 2010;116:2484-2493. doi:10.1182/blood-2010-03-275446 40. Dohner H, Estey E, Grimwade D, et al. Diagnosis and management
29. Zhou Q, Munger ME, Veenstra RG, et al. Coexpression of Tim-3 and of AML in adults: 2017 ELN recommendations from an interna-
PD-1 identifies a CD8+ T-cell exhaustion phenotype in mice with dis- tional expert panel. Blood. 2017;129:424-447. doi:10.1182/blood
seminated acute myelogenous leukemia. Blood. 2011;117:4501-4510. -2016-08-733196
doi:10.1182/blood-2010-10-310425 41. Zhou X, Yao Z, Yang H, Liang N, Zhang X, Zhang F. Are immune-
30. Assi R, Kantarjian H, Ravandi F, Daver N. Immune therapies in related adverse events associated with the efficacy of immune check-
acute myeloid leukemia: a focus on monoclonal antibodies and im- point inhibitors in patients with cancer? A systematic review and
mune checkpoint inhibitors. Curr Opin Hematol. 2018;25:136-145. meta-analysis. BMC Med. 2020;18:87. doi:10.1186/s12916-020-
doi:10.1097/MOH.0000000000000401 01549-2
31. Williams P, Basu S, Garcia-Manero G, et al. The distribution of T-cell 42. Duan J, Cui L, Zhao X, et al. Use of immunotherapy with programmed
subsets and the expression of immune checkpoint receptors and ligands cell death 1 vs programmed cell death ligand 1 inhibitors in patients
in patients with newly diagnosed and relapsed acute myeloid leukemia. with cancer: a systematic review and meta-analysis. JAMA Oncol.
Cancer. 2019;125:1470-1481. doi:10.1002/cncr.31896 2020;6:375-384. doi:10.1001/jamaoncol.2019.5367
32. Daver N, Boddu P, Garcia-Manero G, et al. Hypomethylating agents in 43. Giannopoulos K. Targeting immune signaling checkpoints in acute
combination with immune checkpoint inhibitors in acute myeloid leu- myeloid leukemia. J Clin Med. 2019;8:236. doi:10.3390/jcm8020236
kemia and myelodysplastic syndromes. Leukemia. 2018;32:1094-1105. 44. Latchman Y, Wood CR, Chernova T, et al. PD-L2 is a second ligand
doi:10.1038/s41375-018-0070-8 for PD-1 and inhibits T cell activation. Nat Immunol. 2001;2:261-268.
33. De Sousa Linhares A, Battin C, Jutz S, et al. Therapeutic PD-L1 anti- doi:10.1038/85330
bodies are more effective than PD-1 antibodies in blocking PD-1/PD-L1 45. Okazaki T, Honjo T. PD-1 and PD-1 ligands: from discovery to clini-
signaling. Sci Rep. 2019;9:11472. doi:10.1038/s41598-019-47910-1 cal application. Int Immunol. 2007;19:813-824. doi:10.1093/intimm/
34. Cheson BD, Bennett JM, Kopecky KJ, et al. Revised recommendations dxm057
of the International Working Group for Diagnosis, Standardization 46. Freeman GJ, Long AJ, Iwai Y, et al. Engagement of the PD-1 immu-
of Response Criteria, Treatment Outcomes, and Reporting Standards noinhibitory receptor by a novel B7 family member leads to negative
for Therapeutic Trials in Acute Myeloid Leukemia. J Clin Oncol. regulation of lymphocyte activation. J Exp Med. 2000;192:1027-1034.
2003;21:4642-4649. doi:10.1200/JCO.2003.04.036 doi:10.1084/jem.192.7.1027
35. Thall PF, Sung HG. Some extensions and applications of a Bayesian 47. Philips EA, Garcia-Espana A, Tocheva AS, et al. The structural features
strategy for monitoring multiple outcomes in clinical trials. Stat that distinguish PD-L2 from PD-L1 emerged in placental mammals.
Med. 1998;17:1563-1580. doi:10.1002/(sici)1097-0258(19980 J Biol Chem. 2020;295:4372-4380. doi:10.1074/jbc.AC119.011747
730)17:14<1563::aid-sim873>3.0.co;2-l 48. Yang H, Zhou X, Sun L, Mao Y. Correlation between PD-L2 expres-
36. Chen H, Lau MC, Wong MT, Newell EW, Poidinger M, Chen J. sion and clinical outcome in solid cancer patients: a meta-analysis.
Cytofkit: a Bioconductor package for an integrated mass cytometry data Front Oncol. 2019;9:47. doi:10.3389/fonc.2019.00047

Cancer October 15, 2021

63
TM

Application Note

CyTOF XT
The Next Generation of Mass Cytometry
Introduction
CyTOF Technology. Mass cytometry is a powerful CyTOF Value. Mass cytometers provide impressively
technology that utilizes a time-of-flight mass high resolution of cytometric profiles, empowering both
spectrometer to enable the detection of single cells basic research and practical biomarker-driven clinical
tagged with isotopically pure metal-labeled reagents. research to potentially optimize and personalize disease
CyTOF overcomes the limitations of fluorescence-based management. CyTOF instruments have proven to be
detection systems by separating signals according to valuable additions to cytometry core facilities and service
differences in isotope mass instead of wavelength. labs, and have been adopted for use by key clinical
The negligible signal overlap between masses allows research consortia1,2,3 and in scores of Clinical Research
for simultaneous detection of over 50 targets in a single Trials (see Related documents).
sample tube, a panel size that has not been achieved
by any flow or spectral cytometer to date. The use of a The NeXT Generation. Standard BioTools™ now introduces
universal internal standard—multi-element calibration a new generation of mass cytometer, the CyTOF XT™,
beads—makes it possible to normalize data across (Figure 1). The novel design, fully automated sample
instruments and experiments. Mass cytometry thus acquisition, and easier operational workflows of CyTOF
generates the highest-parameter snapshot of XT simplify the planning and execution of high-parameter
phenotype and function for every cell. cell profiling studies.

Objectives
This application note presents information on key
CyTOF XT features demonstrating the following
advances in instrument benefits and performance:
• Fully automated acquisition with the new
Autosampler to increase productivity and sample
throughput with negligible carryover
• Unique system design and software logic to sense
and remove clogs
• Automated detector voltage (DV) optimization to
maintain signal stability during extended acquisitions
• High degree of agreement between data collected
on CyTOF XT and Helios™ instruments
• An array of data management improvements,
including on-the-fly normalization and processing,
the latest industry file format (FCS 3.1), optimized
storage requirements, a range of troubleshooting
options, and many more

Figure 1. CyTOF XT, featuring a streamlined

design and automatic sample acquisition.
64
CyTOF XT enables walk-away sample Routine workday comparison
acquisition for increased productivity Although many workflow steps for Helios and
CyTOF XT are similar, there are several key differences.
Improving on previous CyTOF instruments that required Figure 2 highlights the significant time savings and
manual loading of each sample before acquisition. streamlined workflow of a model workday using CyTOF
CyTOF XT revolutionizes sample delivery with the XT as compared to a workday with Helios. In this
Autosampler Module and a simplified front end example (Figure 2A), 10 whole blood samples stained
assembly. The new Autosampler consists of 4 major with the Maxpar® Direct™ Immune Profiling Assay™
components: the sample probe, a syringe-based pump (Cat. No. 201325) are acquired in a typical 8-hour
unit, a tray for acquisition and cleaning solutions, and workday. Highlighting the instrument’s advancement
most important, a carousel that holds 13 sample tubes in extended acquisition, Figure 2B demonstrates the
chilled at 4–8 °C. The Autosampler enables automated ability of CyTOF XT to acquire 42 samples within
sample delivery over long acquisitions while maintaining a 23-hour period. The following sections describe
sample integrity. these example workdays on Helios and CyTOF XT
in greater detail.
The CyTOF XT with the Autosampler Module automates
the following processes:
Helios is optimized for single-tube data acquisition.
• Tuning the instrument After the plasma is ignited and the Helios system is
warmed up, tuning solution must be loaded manually,
• Cleaning the sample fluidics followed by instrument tuning, a bead sensitivity test,
and conditioning of the plasma with Maxpar Cell
• Acquisition of samples already in suspension
Acquisition Solution (CAS, Cat. No. 201240). The first
• Resuspension, addition of EQ™ Calibration Beads, sample may then be loaded into the Sample Loader for
and acquisition of pelleted samples acquisition according to the sample criteria selected. In
the workday example (Figure 2A, Helios), purple time
• Detection and removal of clogs bars between samples indicate the manual handling
steps, such as sample preparation (resuspension and
This combination of features greatly reduces the need filtering), data normalization, and cleaning between
for operator interaction with the instrument during samples with CyTOF Washing Solution (Cat. No.
sample acquisition. Hands-on interaction is required 201071) and CAS. In this example, user intervention is
only to load calibration beads and set up samples in the required for routine instrument cleaning and shutdown
carousel (Figure 2). In addition, CyTOF Software v8.0 for procedures at the end of a workday. The Helios
CyTOF XT performs data normalization during sample acquisition of these 10 samples with monitoring requires
acquisition, which further frees up operator time while approximately 7 hr 15 min. (Figure 2A, Helios).
the data file is processed automatically.

A 10 samples acquired within 8 hours B 42 samples acquired within 23 hours

12 12
12

9 3 9 3 9 3
Helios CyTOF XT CyTOF XT

6
6 Hands-on 6
Monitor onsite
Hands-free day
Hands-free night

Figure 2. Comparison of a model workday on Helios and CyTOF XT. A) Representation of Helios (left) and CyTOF XT (right) acquisition of
10 tubes of whole blood samples stained with the Maxpar Direct Immune Profiling Assay. Each purple block represents hands-on time when
interaction with the instrument is required. Orange blocks represent hands-free automated periods. B) Representation of an extended CyTOF
XT acquisition when, upon completion of the first 10 samples, the carousel is reloaded with a new batch of 12 samples stained with the Maxpar
Direct Immune Profiling Assay and 1 large-volume (12 mL) tube of a 20-plex barcoded sample for unattended acquisition. This model example
illustrates how 42 samples could be acquired on the CyTOF XT within a 23-hour period.

65
CyTOF XT reduces hands-on instrument time and CyTOF XT delivers sample batch
increases sample throughput.
acquisitions without sample carryover
CyTOF XT workflow steps are similar to those for Helios,
but most are automated. An operator only needs to The walk-away sample preparation and acquisition
load the bottle tray with required solutions (including enabled by CyTOF XT is achieved through the sample
an improved high-ionic-strength solution, Maxpar handling advancements of the Autosampler. Sample
Cell Acquisition Solution Plus for CyTOF XT, Cat. No. carryover is a concern for any sample introduction system.
201244), ignite and warm up plasma, start the automated This section describes the built-in cleaning of sample
tuning protocol, load the carousel with samples, select probe and Autosampler lines, which minimizes carryover.
acquisition criteria, and walk away while the system
does the rest (Figure 2A, CyTOF XT). CyTOF XT is designed to perform automated washes
of the sample line fluidics. Both the inside and outside
The CyTOF XT automatically adds EQ beads, of the sample probe are washed between samples.
resuspends pelleted samples and mixes prior to Different pre-wash settings can be applied to avoid
acquisition, performs washes, notifies the operator carryover depending on the sample type: Light,
upon successful completion of a batch, normalizes and Medium (default setting), and Heavy, which vary in
saves files, completes extended cleaning after batch washing time, pulsing, and/or solutions used. For
acquisition, and shuts down plasma at the end of a more information, refer to the CyTOF Software v8.0
workday. User attention is needed only for nebulizer Help for CyTOF XT (FLDM-00045), also available as
maintenance cleaning after the instrument shuts the software integrated help guide.
itself down. The carousel is accessible during batch
acquisition, allowing an operator to add new samples or To demonstrate the absence of unwanted sample
replace the acquired tubes with a new batch of samples carryover, the following test was performed.
for continuous acquisition. Finally, the Autosampler
Sample carryover study design
carousel is cooled and kept at 4–8 °C, maintaining
sample integrity for many hours and ensuring high Two separate sets of human peripheral blood
quality data. mononuclear cell (PBMC) samples were barcoded
using 2 different barcoding workflows: Cell-ID™ 20-Plex
In contrast to Helios, the automated sample acquisition Pd Barcoding Kit (Cat. No. 201060) for palladium (Pd)
on CyTOF XT (Figure 2A, orange time bar) frees barcoding up to 20 samples, and 7 individual cadmium
up operator time while instrument automation and (Cd)-labeled anti-CD45 antibodies from Standard BioTools
programming do the work. Acquisition of the 10 assay for 35-plex live-cell barcoding. Barcoding uniquely labels
tubes occurs unattended for 7 hr 50 min on the CyTOF multiple individual samples with a combination of Pd or
XT, enabling the operator to perform other activities Cd isotopes, allowing the samples to be pooled together
through the day. in 1 tube for staining and acquisition. Individual samples
can then be identified based on their unique combination
CyTOF XT extends the workday with of Pd or Cd isotopes. (To learn more about barcoding
walk-away automation. options, refer to The Benefits of Palladium Barcoding on
CyTOF XT provides an opportunity to go beyond the Data Quality and Workflow (FLDM-00012) and Enabling
8-hour workday with sample acquisition for an overnight Live-Cell Barcoding with Anti-CD45 Antibodies in
run. Figure 2B illustrates how the Autosampler and Suspension Mass Cytometry (FLDM-00488) available
CyTOF Software v8.0 for CyTOF XT enable increased at standardbiotools.com.)
sample throughput. An additional 12 samples stained
with the Maxpar Direct Immune Profiling Assay plus Replicates from each barcoded and pelleted sample
a 20-plex barcoded sample may be run unattended were loaded into the CyTOF XT carousel in quadruplicate,
overnight after a standard 8-hour workday on CyTOF XT, in alternating order of Pd- and Cd-barcoded samples.
bringing the total number of samples run to 42. The default pre-wash setting was applied before each
sample. The number of all collected events per tube
Batch sample acquisition is possible because the was approximately 2 x 106 for Pd-barcoded samples and
Autosampler applies stringent washing conditions to 1.5 x 106 for Cd-barcoded samples.
ensure consistent data collection and reduce sample Sample carryover test results and summary
carryover, while detecting and resolving clogs during
acquisition. Refer to the next sections for more about The number of Pd-barcoded events carried over to Cd-
carryover tests and CyTOF XT unclogging capabilities. barcoded samples was negligible. Fewer than 10 events

66
Pd-barcoded Cd-barcoded out of all 600,000 Pd+ live events, or <0.01%
samples samples carryover, were observed (Figure 3). The results
demonstrate that the Autosampler has effective
washing between samples when default settings
are used on the CyTOF XT. The automated pre-wash
Replicate 1

settings can be customized for more stringent

cleaning if required for samples that are anticipated
to be challenging to maintain in single-cell suspension.
<0.01%
Following the clog prevention recommendations
outlined in the next section on challenging sample
handling will also minimize carryover.

CyTOF XT is uniquely designed to

Replicate 2

automatically detect and resolve clogs

Clogging can disrupt and delay cytometry instrument
<0.01% workflow. Helios and CyTOF XT rely on different
strategies to manage and unclog the sample fluidics.
114Cd

Clog management on Helios and CyTOF XT

For challenging samples on the Helios system, a clog
becomes apparent when the sample flow rate drops
Replicate 3

from 30 µL/min to 0 µL/min. The operator must then

perform a manual unclogging procedure.
In contrast to Helios, the CyTOF XT Autosampler
<0.01% software detects potential clogs by monitoring whether
pressure and flow parameters fluctuate beyond set
thresholds. Upon detecting a blockage, the software
executes a routine that mimics the manual unclogging
procedure typically performed by an operator. To
Replicate 4

best demonstrate the improvements made to clog

management and resolution on CyTOF XT, a dissociated
tissue sample test was performed.

<0.01% Dissociated tissue sample study design for CyTOF XT

Multiple replicates of cells from dissociated mouse
intestinal tissue were run for several hours on a CyTOF
105Pd XT at the optimized concentration of 0.5 x 106 cells/mL.
Dissociated cells from solid tissue were chosen because
Figure 3. Negligible Pd-barcoded sample carryover observed in
this sample type is known to be more difficult to maintain
subsequent Cd-barcoded samples. A set of 8 Pd- and Cd-barcoded
samples was run in quadruplicate in alternating order to assess the as a single-cell suspension. Dissociated mouse tissue
carryover of Pd-barcoded cells to the next Cd-barcoded sample. samples were stained according to Maxpar Cell Surface
Default pre-wash settings were applied. Biaxial plots of 105Pd vs. 114Cd Staining with Fresh Fix Protocol (400276) and prepared
show Cd-positive and Pd-positive populations. Samples were gated using validated workflows, which include filtering
on live singlet cell events identified using the cleanup strategy
through 35 µm cell strainers before sample acquisition.
described in Approach to Bivariate Analysis of Data Acquired Using
the Maxpar Direct Immune Profiling Assay (400248). The percentage A batch of 12 dissociated mouse intestinal cell samples
of Pd-barcoded sample carryover is shown in the adjacent Cd- in pelleted format was loaded into the carousel and
barcoded sample. a set volume was acquired. The pressure in the sample
introduction line was monitored and recorded during
sample acquisition. Sample pressure logs and data
quality were analyzed upon completion of the run.

67
Dissociated tissue sample test results and summary Sample 2 and Sample 3 (Figure 4B). Additionally, only
Figure 4A illustrates the pressure changes during the approximately 14% of the expected 700,000 events were
acquisition of the mouse intestinal cell samples using lost due to the 4 clogs in Sample 1 (Figure 4C). More
the default unclogging parameters on CyTOF XT. All important, the automated unclogging routine extended
samples were acquired successfully. As shown in Figure the acquisition by only 30 minutes and completed a
4A, Sample 1 triggered the automated unclogging batch of samples without need for manual intervention.
routine 4 times while Samples 2 and 3 were acquired CyTOF XT is uniquely designed to detect and
without interruptions. Note that despite the occurrence automatically resolve clogs. The combination of
of multiple clogs during the acquisition, marker signal automated features (sample acquisition, unclogging,
intensity was not affected by the automated unclogging and instrument shutdown), makes CyTOF the most
protocol as demonstrated by comparing Sample 1 to productive high-parameter cytometer on the market.

A Sample 1 Sample 2 Sample 3

80
I II III
60

-20
Time, 4.5 hr total
Pressure, psi
B C

Sample 1 Sample 2 Sample 3

Sample 1
Sample 2
Count

Events found 599,673 693,124 719,634

Sample 3
Bead events
29,457 31,540 32,538
found
E-cadherin+ CD45+

Time 1 hr 56 min 1 hr 29 min 1 hr 29 min

Median signal
E-cadherin+ CD45+
intensity Clogs found 4 0 0
Sample 1 49.068 206.416

Sample 2 49.518 207.846

Clogs resolved 4 n/a n/a
Sample 3 49.895 210.411

Figure 4. Automated unclogging by CyTOF XT resolves blockages with no impact on data quality. 12 samples (2.5 mL each) were acquired
automatically with several occurrences of high pressure due to multiple clogs in single samples. A) Pressure monitoring of the 12 samples
identified 3 consecutive samples with varying unclogging routine experiences. Section I (light blue) denotes washing steps between samples.
During this time, rapid pressure swings are observed as positive to negative pressure readings, which function to remove the remaining sample
in the line. Similarly, pressure during the unclogging routine (Section II, light green) fluctuates to effectively remove the clog and enable the
continuation of the acquisition. A rhythmic pattern (Section III, light orange) in the pressure readings is seen in the acquisition of Sample 2 and
Sample 3. This pattern is caused by the change in pressure detected between single loops of a sample that are loaded into the Autosampler
module. The loop accommodates 250 µL of the sample volume and the entire acquisition happens in intervals, hence the rhythmic pressure
pattern. B) The automatic unclogging routine did not impact signal intensity, as illustrated by the unchanged signal intensity of 2 major
populations between all 3 samples. C) Summary of collected events and detected clogs.

68
Tips for success CyTOF XT delivers optimal data quality
Recommendations to reduce clogging: over extended acquisitions.
Sample preparation. Wash samples with Maxpar CAS In some cases, a typical workday of 8 hours may not be
Plus twice immediately prior to loading the carousel. sufficient to fulfill the requirements of the experiment,
Sample filtration. All samples should be filtered as when running samples for a large study. Previous
through appropriately sized cell strainers to prevent sections demonstrated that the CyTOF XT is uniquely
cell aggregates and large particles from entering the designed to accommodate long unattended runs
fluidics. Best practice is not to force the sample through and can successfully resolve clogs and avoid sample
the mesh with a pipette but instead gently pipette the carryover. The following section assesses the CyTOF
sample through the mesh cap. Avoid using a pipette to XT detector’s potential to maintain a high level of
aspirate any remaining sample from the inner side of the performance during long hours of ion detection.
straining cap. A new feature of CyTOF XT is the automatic adjustment
Cell concentration. Optimize cell concentration based of the detector voltage. This ensures signal stability
on sample type. Acquire potentially challenging samples regardless of acquisition time.
at 0.5 x 106 cells/mL or less.
Extended acquisition study design
Pre-wash settings. Select a heavy pre-wash cycle for
The goal of this study was to confirm that all marker
challenging samples.
signal intensities were maintained at the same level
Acquisition volume. Split large volume samples into and did not decrease over time due to any loss of
smaller volume acquisitions over multiple tubes. sensitivity by the CyTOF XT detector. Twelve tubes
Regular instrument maintenance is key. Clean and containing various samples were loaded into the
maintain parts and fluidics as per Standard BioTools carousel for long-term unsupervised acquisition.
recommendation. For more detail, refer to the CyTOF XT To assess the impact of time on data quality, the first
User Guide (FLDM-00254) and the CyTOF Software v8.0 and last tube of the carousel were loaded with the
Help for CyTOF XT (FLDM-00045), also available as the same sample. This sample was prepared by staining
software integrated help guide. PBMC with surface and nuclear markers according
to the Maxpar Nuclear Antigen Staining with Fresh
Detailed recommendations for reducing clogs can also Fix Protocol (400277). The time difference between
be found in the CyTOF Software v8.0 Help for CyTOF XT acquisition of the sample in Run 1 and Run 12 was
(FLDM-00045). approximately 19 hours. During this time, the chilled
carousel maintained the samples in suspension at
4–8 °C. Test results are presented in Figure 5.

Figure 5. Signal stability over 19 hours of acquisition. 12 tubes of various sample types were continuously collected on CyTOF XT over 19
hours. Run 1 and Run 12 are replicates of the same sample and were acquired first and last, respectively. Run 12 and Run 1 demonstrated the same
level of signal. 14 markers are shown as representative data in order of their cell marker signal intensities (low to high). Numbers represent signal
intensities (median) over the 19-hour period.

69
Extended acquisition study results and summary PBMC sample in Experiment 1 and a single donor whole
Fourteen surface markers of low, medium, and high blood sample in Experiment 2.
expression range were analyzed. The absolute Experiment 1: PBMC study design
difference in median signal intensity for each marker
was calculated between Run 1 and Run 12. Low PBMC from a single donor were stained with the Maxpar
expression markers with signal intensity <50 dual Direct Immune Profiling Assay and run in triplicate on
counts (CD38, CD25, CD20, CD127, CD19, CD56) both CyTOF XT and Helios in parallel. PBMC sample
showed an average difference of only 4.9%. Medium preparation and acquisition were performed according
expression markers (≥50 and 200 dual counts: CD11b, to recommendations in the Maxpar Direct Immune
CD3, HLA-DR, CD4, CD45) exhibited an average Profiling Assay Cell Staining and Data Acquisition User
difference of 1.9%. High expression markers with signal Guide (400286), and the Helios User Guide (400250)
of >200 counts (CD8, CD14, DNA) demonstrated a or the CyTOF Software v8.0 Help for CyTOF XT
negligible average difference of 1.2%. (FLDM-00045), also available as the software integrated
help guide. Data was normalized using the applicable
This study demonstrates that the automated detector CyTOF Software and analyzed with Maxpar Pathsetter™,
voltage optimization of CyTOF XT can sustain a constant a fully automated data analysis solution for samples
level of instrument sensitivity during extended runs. processed with the Maxpar Direct Immune Profiling
Comparable data performance on Assay. Maxpar Pathsetter generates a report with
frequencies of defined populations and their signal
Helios and CyTOF XT intensities. As a comparative analysis, each population
Successful completion of extended runs is facilitated was gated manually in Cytobank and compared
by CyTOF Software v8.0 for CyTOF XT, which has been between the instruments. The manual gating strategies
optimized to improve the experience of acquiring were applied following the technical note: Approach to
samples, data processing, and high-parameter data Bivariate Analysis of Data Acquired Using the Maxpar
storage. The enhancements include the following: Direct Immune Profiling Assay (400248).
• CyTOF XT software automates detector voltage Experiment 2: Whole blood study design
optimization during acquisition, maintaining
A test similar to Experiment 1 was performed with whole
comparable sensitivity throughout extended
blood stained with the Maxpar Direct Immune Profiling
acquisitions.
Assay and run in triplicate on both CyTOF XT and Helios
• The software improves upon the built-in sample in parallel. Data were normalized and analyzed as
standardization, using EQ Six Element or EQ Four described in Experiment 1.
Element Calibration Beads (Cat. Nos. 201245 and
Comparison of signal intensity and population
201078, respectively) to minimize technical variability.
frequency
The sample normalization algorithm better identifies
the EQ calibration beads on the fly, delivering ready- The automated Maxpar Pathsetter analysis compiled
to-analyze data immediately after acquisition. cell classification data from the triplicate PBMC and
whole blood datasets acquired on Helios or CyTOF
• CyTOF XT software records a linear mode data
XT. The results of 35 cell populations were graphed
(LMD) file, which captures all 135 channels for post-
in the upper panel of Figures 6 and 7 for PBMC and
acquisition reprocessing and troubleshooting.
whole blood samples, respectively. Comparison of
Both unprocessed and normalized flow cytometry
population frequencies of data files from Helios and
standard (FCS) files are recorded using FCS 3.1
CyTOF XT automatically analyzed in Maxpar Pathsetter
format. The final output files have been optimized for
demonstrated comparable results for replicate samples
more efficient use of data storage.
between the two instruments (Figures 6 and 7, upper
These CyTOF XT software advancements significantly panels). The percent difference was <3.2% and <17.4%
enhance data processing and workflows while for major (>20% of all live events) and minor (<20%
maintaining comparable performance between Helios of all live events) populations, respectively, in both
and CyTOF XT. This is illustrated in the section below. experiments (Appendix, Tables S1 and S4). The lower
Study design to compare signal intensities and panels of Figures 6 and 7 include examples of signal
population frequencies on Helios and CyTOF XT intensities for major populations that were manually
gated. The median intensities of positive populations
Two independent experiments were performed to
were comparable between Helios and CyTOF XT by
demonstrate the comparability of data with regards
manual gating (data not shown).
to signal intensities and cell population frequencies
between CyTOF XT and Helios data. Tests were run on Using this dataset, an assessment of staining quality for
CyTOF XT and Helios in parallel using a single donor each marker on CyTOF XT and Helios was conducted.

70
Maxpar Pathsetter performs a staining assessment using Deming regression was also used to statistically
a statistical approach called strictly standardized mean compare population frequencies, median signal
difference (SSMD or β). SSMD considers the median and intensities, and β values for both PBMC and whole blood
median absolute deviations of both positive and negative tests performed on Helios and CyTOF XT (Appendix,
populations to assess the resolution of each marker, Figures S1–S4). There was no significant difference
known as a β value. A higher β value indicates a higher between these measurements on each instrument
marker resolution that translates to a better separation except in the β values of the whole blood tests, which
of positive and negative populations. For the PBMC were higher on CyTOF XT. Deming regression analysis
and whole blood tests, the β value demonstrated a found statistically higher B values for samples acquired
high level of reproducibility across CyTOF XT and on CyTOF XT relative to Helios. (Appendix, Figure S4).
Helios data (Appendix, Tables S2 and S6). Furthermore,
β values for CyTOF XT data generally exceeded those
of Helios.

50 10 3
Population frequency

40
Helios
CyTOF XT
(percentage)

2
30
5
20
1

0 0 0

sin ils

In hils
CM

ut sts
Th ike
Th like

CD reg
4 ells

4 lls

Gr me T

so s
B e lls

CD EM

ul ory
L a ells

CD TE

Th TE

m s
K

o
CD NK

AI ke

C
B lls

8 M
CD D4 e
8 CM

NK ono

CD ive
m s

m δT
3 tes

on ive

Ba phil
NC yte
B /N K
as yte

on
on
C aiv

pD
D
N

CD 8 E

Eo ph
CD c e

M -l i

Ne bla
na

1-l

op
T
8

an m
4

as γ
CD T c

c
M na

2-
te
CD cy

tm
4

8
Cl o c
n

ro
T
T

a
o

Ea
s
ph
m

Pl
Ly

Figure 6. Comparison of PBMC cell population frequencies and signal intensities between Helios and CyTOF XT 6 replicates of PBMC from
the same donor were stained with the Maxpar Direct Immune Profiling Assay, pooled together, and split into 6 tubes for acquisition on Helios and
CyTOF XT, with 300,000 events acquired per sample. Automated population frequency data obtained from Maxpar Pathsetter analysis is
presented in the bar charts (upper panel). Biaxial plots of major populations (lower panel) are used to compare signal intensity (median) or
percentage of events as specified in each corresponding plot between Helios and CyTOF XT by manual gating.

71
Population frequency 60 10 6
Helios
50 5
CyTOF XT 8
(percentage)

40 4
6
30 3
4
20 2

2
10 1

0 0 0

B lls

Ea lls
B e

K
NK ils
CD M

CD M

sin no
CD TE

L a NK
CD M

ive

C
Th like
Th KT
Th T

8 ve

ab y

m s
T/ g

so o
ils
Pl me ke
CD ive

o
In DC
4 ells
8 ells
L y utr tes

lls
o ls

as n

m or
N

NC last
3 tes

M Tre

Ba on

on
E
C

D
ce
h
ce
γδ

ph
Eo mo
ph hi

Cl Mo

B 7-l i
CD nai
ce

na
1-l

N
8

na
4

p
te
rly

as m
Ne loc y

2-
4

tm
m op

CD T c
CD T c

8
CD cy

1
T

AI
CD
a nu
Gr

Figure 7. Comparison of stained whole blood cell population frequencies and signal intensities between Helios and CyTOF XT. 6 replicates
of whole blood from the same donor were stained with the Maxpar Direct Immune Profiling Assay, pooled together, and split into 6 tubes for
acquisition on Helios and CyTOF XT, with 400,000 events acquired per sample. Automated population frequency data obtained from Maxpar
Pathsetter analysis is presented in the bar charts (upper panel). Biaxial plots of major populations (lower panel) are used to compare signal
intensity (median) and percentage of events as specified in each corresponding plot between Helios and CyTOF XT by manual gating.

CyTOF XT can resolve rare populations. cell population in whole blood. As shown in the far-
The high-quality data delivered by CyTOF XT enables right biaxial plot with fixed gates in Figure 8, the same
the identification of both abundant and rare populations sample acquired on Helios and CyTOF XT yielded a
in a sample. An example step-by-step manual gating similar number of MAIT/NKT CD4– cells (1,047 and 1,055,
strategy is shown in Figure 8, demonstrating that 2 respectively) out of 400,000 total events collected.
replicates of the same sample acquired on both In this section we showed that comparable performance
instruments reproducibly resolve markers to identify between Helios and CyTOF XT is demonstrated
a desired cell population. The MAIT/NKT CD4– through the analysis of signal intensities, cell population
cell population was chosen because it is a rare frequencies, and marker resolution of samples stained
population that comprises only 0.4% of the total live with a high parameter-marker panel.

72
Figure 8. Manual gating strategy of the rare MAIT/NKT CD4– T cell populations, acquired from the same whole blood sample on both
Helios and CyTOF XT. An example step-by-step gating strategy for a given rare population after applying the cleanup strategy and gating outlined
in Approach to Bivariate Analysis of Data Acquired Using the Maxpar Direct Immune Profiling Assay (400248) is presented. All 35 populations can
be identified both manually and automatically using normalized files from both instruments. As shown in the CD28/CD161
gate, the total number of MAIT/NKT CD4– T cells is highly comparable between Helios and CyTOF XT.

Summary Conclusion
CyTOF XT is a new generation of CyTOF instrument that Mass cytometry is widely recognized as a technology
shares the same reliable level of performance with its that brings a level of multiplexing, precision, and
predecessor, Helios, but with several key improvements: reproducibility to cell analysis not enabled by other
• The novel Autosampler design and new CyTOF single-cell platforms.
Software v8.0 for CyTOF XT features automated CyTOF XT further enhances the capabilities of mass
sample acquisition, EQ bead addition to samples, cytometry with workflow automation that refines and
unclogging, normalization, cleaning, and plasma simplifies sample processing and data acquisition.
shutdown. Together, these features improve This next generation CyTOF instrument offers a
workflows to reduce hands-on time without new level of autonomy and reproducibility through
impacting data quality. streamlined operation, automated system monitoring,
• Batch acquisition enables up to 13 sample tubes and easier system maintenance, making CyTOF XT the
to be loaded into the chilled carousel of the superior choice for high-parameter cytometric analysis
Autosampler module, with the added option in clinical and translational research studies.
to add more samples upon completion of earlier
tubes, for added productivity and flexibility. References
• An onboard bottle tray and a chilled Autosampler
carousel facilitate extended acquisitions. 1 Chen, H.X., Song, M., Maecker, H.T. et al. “Network for
biomarker immunoprofiling for cancer immunotherapy:
• Robust automated pre-wash cycles minimize Cancer Immune Monitoring and Analysis Centers and
sample carryover. Cancer Immunologic Data Commons (CIMAC-CIDC).”
• Automated detector voltage optimization function Clinical Cancer Research (2021): doi:10.1158/1078-0432.
helps to maintain consistent signal intensities over CCR-20-3241.
the course of extended acquisitions. 2 Accelerating Partnership Detailed Research Plan:
• On-the-fly data processing and normalization with RA, SLE and Related Autoimmune Disorders
the latest FCS format (version FCS 3.1) provides Steering Committee. National Institute of Arthritis and
faster time to results with smaller file sizes. Musculoskeletal Diseases.
3 Guo, N., van Unen, V., Ijsselsteijn, M.E. et al. “A
34-marker panel for imaging mass cytometric analysis
of human snap-frozen tissue.” Frontiers in Immunology
11 (2020): 1466. doi:10.3389/fimmu.2020.01466.

73
Related documents • Maxpar Cell Surface Staining with Fresh Fix
Protocol (400276)
Go to standardbiotools.com and search for the following
related documents. • Maxpar Nuclear Antigen Staining with Fresh Fix
Protocol (400277)
• Use of CyTOF Technology in Clinical Research Trials
Data Sheet Appendix:
• The Benefits of Palladium Barcoding on Data Quality Supplemental material
and Workflow Application Note (FLDM-00012)
Data analysis of population frequencies and
• Enabling Live-Cell Barcoding with Anti-CD45 signal intensities
Antibodies in Suspension Mass Cytometry
The following section details additional analysis of
Application Note (FLDM-00488)
files described in the application note. Two sets of
• CyTOF XT User Guide (FLDM-00254) tables and figures are provided for live singlet events
• Helios, a CyTOF System User Guide (400250) of PBMC and whole blood. Samples were acquired
in triplicate on both Helios and CyTOF XT. Raw FCS
• Maxpar Direct Immune Profiling Assay Cell Staining data were normalized in CyTOF Software and then
and Data Acquisition User Guide (400286) analyzed in Maxpar Pathsetter for summary statistics
• CyTOF Software v8.0 Help for CyTOF XT (FLDM-00045) and calculations of β values. Databases from Maxpar
• Approach to Bivariate Analysis of Data Acquired Pathsetter were extracted and further analyzed in NCSS
Using the Maxpar Direct Immune Profiling Assay Statistical Analysis and Graphics software program for
Technical Note (400248) Deming regression.

Population % live, Helios and CyTOF XT (PBMC samples)

Mean, standard deviation (SD), and percent coefficient of variation (%CV) of population frequencies of PBMC samples stained using
Descriptive
the Maxpar Directstatistic
Immune Profiling Assay from the same donor prepared in individual assay tubes, pooled together, and split into 2 sets of
triplicates, 1 triplicate set per instrument

Helios CyTOF XT
Population % Live Mean SD % CV Population % Live Mean SD % CV
1 Lymphocytes 41.58 0.99 2.38 1 Lymphocytes 41.33 0.18 0.44
2 CD3 T cells 26.35 0.77 2.92 2 CD3 T cells 26.31 0.28 1.08
3 CD8 T cells 5.61 0.42 7.46 3 CD8 T cells 5.74 0.03 0.47
4 CD8 naive 0.72 0.10 13.26 4 CD8 naive 0.74 0.02 2.66
5 CD8 central memory 0.50 0.07 14.25 5 CD8 central memory 0.44 0.04 8.62
6 CD8 effector memory 2.50 0.12 4.93 6 CD8 effector memory 2.55 0.02 0.73
7 CD8 terminal effector 1.89 0.19 10.01 7 CD8 terminal effector 2.01 0.05 2.65
8 CD4 T cells 20.09 0.38 1.89 8 CD4 T cells 20.03 0.15 0.74
9 CD4 naive 2.23 0.01 0.47 9 CD4 naive 2.27 0.06 2.54
10 CD4 central memory 3.72 0.12 3.11 10 CD4 central memory 3.76 0.15 4.01
11 CD4 effector memory 5.12 0.22 4.29 11 CD4 effector memory 4.99 0.25 5.03
12 CD4 terminal effector 9.02 0.26 2.84 12 CD4 terminal effector 9.01 0.11 1.20
13 γδ T cells 0.42 0.02 3.86 13 γδ T cells 0.35 0.12 34.37
14 MAIT/NKT 0.23 0.02 7.58 14 MAIT/NKT 0.19 0.00 0.20
15 B cells 5.77 0.25 4.30 15 B cells 5.79 0.05 0.87
16 B naive 4.18 0.23 5.57 16 B naive 4.17 0.06 1.33
17 B memory 1.57 0.02 1.32 17 B memory 1.59 0.04 2.47
18 Plasmablasts 0.03 0.01 38.11 18 Plasmablasts 0.03 0.00 11.57
19 Natural killer cells 9.45 0.06 0.68 19 Natural klller cells 9.23 0.15 1.64
20 Early natural killer 4.67 0.03 0.70 20 Early natural killer 4.50 0.04 0.91
21 Late natural killer 4.78 0.03 0.67 21 Late natural killer 4.73 0.12 2.46
22 Monocytes 44.72 0.66 1.48 22 Monocytes 43.52 0.35 0.80
23 Classical monocytes 43.28 0.73 1.69 23 Classical monocytes 42.04 0.32 0.76
24 Intermediate monocytes 1.20 0.07 6.11 24 Intermediate monocytes 1.26 0.04 3.03
25 Non-classical monocytes 0.24 0.02 6.70 25 Non-classical monocytes 0.22 0.01 6.06
26 Plasmacytoid dendritic cells 0.21 0.00 1.58 26 Plasmacytoid dendritic cells 0.21 0.02 11.34
27 Myeloid dendritic cells 1.06 0.03 2.70 27 Myeloid dendritic cells 1.04 0.04 3.95
28 Granulocytes 4.24 0.10 2.37 28 Granulocytes 4.48 0.12 2.64
29 Neutrophils 0.59 0.06 10.46 29 Neutrophils 0.73 0.04 5.94
30 Basophils 2.46 0.04 1.65 30 Basophils 2.50 0.03 1.36
31 Eosinophils 0.05 0.00 8.69 31 Eosinophils 0.06 0.01 16.86
32 Treg 0.43 0.01 2.78 32 Treg 0.41 0.02 4.39
33 Th1-like 0.78 0.00 0.30 33 Th1-like 0.86 0.01 1.71
34 Th2-like 3.02 0.09 3.03 34 Th2-like 3.02 0.03 0.88
35 Th17-like 0.35 0.04 11.94 35 Th17-like 0.37 0.04 10.39

Table S1. Descriptive statistics of population frequencies as percentage of total live singlet cells for 1 triplicate from Helios and CyTOF XT

74
Population
Population % live,
Population % live,and
Helios
frequencies Helios and
for CyTOF
each CyTOF XT stained with the Maxpar Direct Immune Profiling Assay from the same donor prepared in
XT replicate
PBMC
(PBMC
(PBMC samples)
individual samples)
assay tubes, pooled together, and split into 2 sets of triplicates, 1 triplicate set per instrument
Deming regression
Deming regression
% of All Live Helios Helios Helios CyTOF XT CyTOF XT CyTOF XT
% of All Live Helios Helios
Replicate 1 Helios
Replicate 2 CyTOF XT 3
Replicate CyTOF XT 1
Replicate CyTOF XT 2
Replicate Replicate 3
Replicate 1 Replicate 2 Replicate 3 Replicate 1 Replicate 2 Replicate 3
Lymphocytes 41.97 40.45 42.31 41.41 41.12 41.46
Lymphocytes
CD3 41.97
T cells 40.45
26.71
42.31
25.47
41.41
26.88
41.12
26.24
41.46
26.06 26.62
CD3 T cells
CD826.71
T cells 25.47
5.85
26.88
5.12
26.24
5.85
26.06
5.75
26.62
5.71 5.75
CD8 T cells CD85.85
naive 5.12 5.85 5.75 5.71 5.75
0.82 0.63 0.73 0.72 0.76 0.73
CD8 naive CD80.82
central memory 0.63 0.73 0.72 0.76 0.73
0.59 0.45 0.48 0.42 0.41 0.48
CD8 central memory CD80.59
effector memory0.45 0.48 0.42 0.41 0.48
2.49 2.38 2.62 2.54 2.57 2.55
CD8 effector memoryCD82.49
terminal effector2.381.96 2.62
1.67 2.54
2.03 2.57
2.07 2.55
1.97 2.00
CD8 terminal effector
CD4 1.96
T cells 1.6720.19 2.03
19.67 2.07
20.41 1.97
19.95 2.00
19.94 20.20
CD4 T cells CD4 20.19
naive 19.67
2.23 20.41
2.22 19.95
2.24 19.94
2.31 20.20
2.20 2.30
CD4 naive CD4 2.23
central memory 2.22 3.76 2.24
3.81 2.31
3.59 2.20
3.87 2.30
3.81 3.59
CD4 central memoryCD4 3.76
effector memory3.815.18 3.59
4.87 3.87
5.30 3.81
4.88 3.59
4.82 5.28
CD4 effector memory
CD4 5.18
terminal effector4.87
9.01 5.30
8.76 4.88
9.27 4.82
8.89 5.28
9.10 9.04
CD4 terminal effector
γδ T 9.01
cells 8.76
0.43 9.27
0.43 8.89
0.40 9.10
0.35 9.04
0.23 0.47
γδ T cells MAIT/NKT
0.43 0.43
0.24 0.40
0.24 0.35
0.21 0.23
0.19 0.47
0.19 0.19

Helios
MAIT/NKT B cells
0.24 0.24
5.88 0.21
5.49 0.19
5.95 0.19
5.84 0.19
5.74 5.78

Helios
B cells B naive
5.88 5.49
4.31 5.95
3.91 5.84
4.32 5.74
4.19 5.78
4.10 4.20
B naive B memory
4.31 3.911.55 1.56
4.32 1.59
4.19 1.61
4.10 1.61
4.20 1.54
B memory Plasmablasts
1.55 0.02
1.56 0.02
1.59 0.04
1.61 0.04
1.61 0.03
1.54 0.04
Plasmablasts Natural
0.02killer cells 0.02 9.38 9.49
0.04 9.48
0.04 9.32
0.03 9.32
0.04 9.06
Natural killer cells Early9.38
natural killer 9.49 4.63 4.69
9.48 4.69
9.32 4.54
9.32 4.51
9.06 4.46
Early natural killer Late 4.63
natural killer 4.74
4.69 4.80
4.69 4.80
4.54 4.78
4.51 4.81
4.46 4.60
Late natural killer Monocytes
4.74 44.23
4.80 45.48
4.80 44.46
4.78 43.47
4.81 43.90
4.60 43.20
Monocytes Classical
44.23monocytes 45.48 42.71 44.11
44.46 43.03
43.47 42.02
43.90 42.37
43.20 41.73
Intermediate
Classical monocytes 42.71 mono 1.27
44.11 1.12
43.03 1.21
42.02 1.25
42.37 1.31
41.73 1.24
Intermediate mono Non-classical
1.27 mono 1.120.25 1.210.25 0.22
1.25 0.21
1.31 0.22
1.24 0.23
Non-classical monoPlasmacytoid
0.25 dendritic 0.20
0.25 0.20
0.22 0.21
0.21 0.23
0.22 0.22
0.23 0.18
Myeloid
Plasmacytoid dendritic 0.20dendritic 1.05
0.20 1.05
0.21 1.10
0.23 1.07
0.22 1.06
0.18 1.00
Granulocytes 4.33 4.26 4.13 4.49 4.36 4.60
Myeloid dendritic 1.05 1.05 1.10 1.07 1.06 1.00
Neutrophils 0.63 0.52 0.62 0.78 0.73 0.69
Granulocytes 4.33 4.26 4.13 4.49 4.36 4.60
Basophils 2.49 2.49 2.42 2.51 2.54 2.47
Neutrophils 0.63 0.52 0.62 0.78 0.73 0.69
Eosinophils 0.05 0.06 0.05 0.05 0.07 0.05
Basophils 2.49 2.49 2.42 2.51 2.54 2.47
Treg 0.43 0.41 0.44 0.40 0.44 0.40
Eosinophils 0.05 0.06 0.05 0.05 0.07 0.05
Th1-like 0.78 0.78 0.77 0.85 0.88 0.86
Treg 0.43 0.41 0.44 0.40 0.44 0.40
Th2-like 3.10 2.92 3.03 3.01 3.01 3.05
Th1-like 0.78 0.78 0.77 0.85 0.88 0.86
Th17-like 0.30 0.37 0.38 0.33 0.39 0.40
Th2-like 3.10 2.92 3.03 3.01 3.01 3.05
Th17-like 0.30 0.37 0.38 0.33 0.39 0.40
CyTOF XT
Table S2. Raw values of population frequencies as percentage of total
CyTOF XT
Figure S1. Deming regression of population frequencies between the
live singlet cells for 2 sets of triplicates from Helios and CyTOF XT. The 2 sets of triplicate data from Table S2. Null hypothesis is accepted at a
numbers were extracted from Maxpar Pathsetter analysis results. significance level of 0.05, indicating high similarity between the 2 datasets.

Beta values of marker intensities of each PBMC replicate stained with Maxpar Direct Immune Profiling Assay from the same donor on
both instruments
Helios Helios Helios CyTOF XT CyTOF XT CyTOF XT
Helios Replicate 1Helios
Helios Replicate 2CyTOF
Replicate
XT 3 CyTOF
Replicate
XT 1 CyTOF
Replicate
XT 2 Replicate 3
Replicate 1 Replicate 2 Replicate 3 Replicate 1 Replicate 2 Replicate 3
CD38 4.15 4.45 4.23 3.63 3.52 3.42
CD38 4.15 CD14 4.455.39 4.235.41 3.63
5.80 3.52
5.29 3.42
5.55 5.35
CD14 5.39CD19 5.414.17 5.80
4.33 5.29
4.27 5.55
4.42 5.35
4.51 4.39
CD19 4.17 CD3 4.334.35 4.274.43 4.42
4.40 4.51
4.39 4.39
4.49 4.57
CD3 4.35CD45RA 4.432.57 4.402.43 2.57
4.39 2.26
4.49 2.68
4.57 2.35
CD45RA 2.57CXCR3 2.430.52 2.570.53 0.52
2.26 0.46
2.68 0.49
2.35 0.48
CXCR3 0.52CXCR5 0.531.04 1.02
0.52 1.03
0.46 1.02
0.49 1.03
0.48 1.03
CXCR5 1.04 CCR4 1.021.48 1.031.72 1.022.04 1.37
1.03 1.75
1.03 2.40

CCR4 1.48 TCRgd 1.721.83 1.91

2.04 1.372.21 2.49
1.75 2.77
2.40 1.76

TCRgd 1.83 CD28 1.91 3.51 2.213.52 3.22

2.49 3.58
2.77 3.38
1.76 3.35

CD28 3.51 CD127 3.522.97 3.222.94 2.98

3.58 2.89
3.38 3.03
3.35 2.97

CD127 2.97CD56 2.943.54 2.983.62 3.56

2.89 3.71
3.03 3.76
2.97 3.68

CD56 3.54CD161 3.623.09 3.563.09 3.713.85 3.94

3.76 3.70
3.68 4.16
Helios

CD161 3.09CD8 3.093.27 3.853.52 3.943.25 3.13

3.70 3.20
4.16 3.12
Helios

CD8 3.27CD4 3.525.38 3.255.48 3.135.36 5.51

3.20 5.56
3.12 5.55
CCR7 2.41 2.83 2.84 2.72 2.82 2.86
CD4 5.38 5.48 5.36 5.51 5.56 5.55
CD25 2.66 2.55 2.57 2.82 2.65 2.68
CCR7 2.41 2.83 2.84 2.72 2.82 2.86
HLADR 3.15 3.10 3.14 3.33 3.22 3.16
CD25 2.66 2.55 2.57 2.82 2.65 2.68
CD20 3.16 3.23 3.14 3.38 3.35 3.17
HLADR 3.15 3.10 3.14 3.33 3.22 3.16
IgD 2.71 2.80 2.63 2.84 2.83 2.80
CD20 3.16 3.23 3.14 3.38 3.35 3.17
CD57 2.84 2.80 2.92 2.81 2.83 2.69
IgD 2.71 2.80 2.63 2.84 2.83 2.80
CD66b 3.70 3.61 3.62 3.85 3.78 3.39
CD57 2.84 2.80 2.92 2.81 2.83 2.69
CD123 4.01 3.94 4.04 3.97 3.98 4.00
CD66b 3.70 3.61 3.62 3.85 3.78 3.39
CD11c 6.66 6.98 6.73 7.31 7.00 6.87
CD123 4.01 3.94 4.04 3.97 3.98 4.00
CD27 2.61 2.91 2.80 2.91 2.93 2.95
CD11c 6.66 6.98 6.73 7.31 7.00 6.87
CD45 1.63 1.93 2.08 1.80 1.66 1.92
CD27 2.61 2.91 2.80 2.91 2.93 2.95
CCR6 1.65 1.63 1.63 1.63 1.61 1.57
CD45 1.63 1.93 2.08 1.80 1.66 1.92
CD294 3.82 3.70 3.78 3.08 2.93 2.75
CCR6 1.65 1.63 1.63 1.63 1.61 1.57
CD16 1.83 1.79 1.86 1.75 1.85 1.74
CD294 3.82 3.70 3.78 3.08 2.93 2.75
CD45RO 1.57 1.54 1.64 1.41 1.58 1.47
CD16
CD45RO
1.83
1.57
1.79
1.54
1.86
1.64
1.75
1.41
1.85
1.58
1.74
1.47
CyTOF XT
CyTOF XT
Table S3. Beta (β) values of marker intensities of each replicate from Figure S2. Deming regression of β values of marker intensities from
Helios and CyTOF XT. The numbers were extracted from Maxpar Table S3. Null hypothesis is accepted at a significance level of 0.05,
Pathsetter analysis results. indicating high similarity between the 2 datasets.

75
Population % live, Helios and
CyTOF XT (whole blood samples)
Mean, SD, and %CV of population frequencies of whole blood samples stained with the Maxpar Direct Immune Profiling Assay from the
Descriptive statistic
same donor prepared in individual assay tubes, pooled together, and split into 2 sets of triplicates, 1 triplicate set per instrument

Helios CyTOF XT
Population % Live Mean SD % CV Population % Live Mean SD % CV
1 Lymphocytes 33.34 0.57 1.71 1 Lymphocytes 33.44 1.07 3.20
2 CD3 T cells 23.84 0.47 1.96 2 CD3 T cells 23.84 0.87 3.65
3 CD8 T cells 7.73 0.12 1.56 3 CD8 T cells 7.72 0.27 3.47
4 CD8 naive 3.67 0.07 1.83 4 CD8 naive 3.76 0.14 3.77
5 CD8 central memory 0.73 0.02 3.22 5 CD8 central memory 0.81 0.07 8.14
6 CD8 effector memory 1.68 0.05 2.89 6 CD8 effector memory 1.58 0.04 2.71
7 CD8 terminal effector 1.65 0.04 2.34 7 CD8 terminal effector 1.57 0.03 1.76
8 CD4 T cells 15.02 0.32 2.12 8 CD4 T cells 15.03 0.56 3.75
9 CD4 naive 4.24 0.14 3.36 9 CD4 naive 4.20 0.17 4.10
10 CD4 central memory 6.34 0.24 3.72 10 CD4 central memory 6.78 0.34 4.98
11 CD4 effector memory 3.23 0.27 8.40 11 CD4 effector memory 2.93 0.03 0.90
12 CD4 terminal effector 1.20 0.02 1.69 12 CD4 terminal effector 1.12 0.11 9.72
13 γδ T cells 0.68 0.03 3.94 13 γδ T cells 0.68 0.02 2.74
14 MAIT/NKT 0.42 0.01 2.17 14 MAIT/NKT 0.41 0.03 6.16
15 B cells 5.97 0.16 2.62 15 B cells 6.16 0.04 0.69
16 B naive 4.56 0.13 2.82 16 B naive 4.63 0.02 0.48
17 B memory 1.37 0.04 2.66 17 B memory 1.49 0.04 2.57
18 Plasmablasts 0.04 0.01 15.40 18 Plasmablasts 0.04 0.00 9.51
19 Natural killer cells 3.53 0.10 2.97 19 Natural killer cells 3.45 0.19 5.48
20 Early natural killer 1.77 0.04 2.49 20 Early natural killer 1.72 0.12 6.80
21 Late natural killer 1.76 0.06 3.47 21 Late natural killer 1.73 0.07 4.28
22 Monocytes 6.35 0.16 2.59 22 Monocytes 6.52 0.20 3.09
23 Classical monocytes 4.79 0.10 1.99 23 Classical monocytes 5.03 0.10 2.04
24 Intermediate monocytes 0.53 0.02 4.55 24 Intermediate monocytes 0.47 0.04 8.89
25 Non-classical monocytes 1.04 0.05 4.72 25 Non-classical monocytes 1.02 0.06 5.80
26 Plasmacytoid dendritic cells 0.10 0.00 2.46 26 Plasmacytoid dendritic cells 0.09 0.00 5.04
27 Myeloid dendritic cells 0.24 0.01 5.81 27 Myeloid dendritic cells 0.25 0.02 8.88
28 Granulocytes 52.39 0.83 1.59 28 Granulocytes 51.95 1.42 2.73
29 Neutrophils 48.19 0.76 1.57 29 Neutrophils 48.04 1.18 2.46
30 Basophils 0.61 0.02 3.40 30 Basophils 0.63 0.02 3.79
31 Eosinophils 2.86 0.11 3.84 31 Eosinophils 2.46 0.22 8.98
32 Treg 0.62 0.03 5.20 32 Treg 0.57 0.02 4.20
33 Th1-like 0.94 0.03 3.38 33 Th1-like 0.86 0.06 7.12
34 Th2-like 1.03 0.05 4.90 34 Th2-like 1.04 0.03 2.98
35 Th17-like 1.34 0.02 1.68 35 Th17-like 1.39 0.06 4.22

Tables S4. Descriptive statistics of population frequencies as percentage of total live singlet cells for triplicate sample data from Helios and CyTOF XT

76
Population frequencies of each whole blood sample stained with the Maxpar Direct Immune Profiling Assay from the same donor prepared in
individual assay tubes, pooled together, and split into 2 sets of triplicates, 1 triplicate set per instrument

Helios Helios Helios CyTOF XT CyTOF XT CyTOF XT

Replicate 1 Helios2
Replicate Helios 3
Replicate Helios 1
Replicate CyTOF XT
Replicate 2 CyTOF XT3
Replicate CyTOF XT
Replicate 1 Replicate 2 Replicate 3 Replicate 1 Replicate 2 Replicate 3
Lymphocytes 32.90 33.98 33.13 34.66 33.01 32.66
Lymphocytes 32.90 33.98 33.13 34.66 33.01 32.66
CD3 T cells
CD323.53
T cells 24.37
23.53
23.61
24.37
24.81
23.61
23.57
24.81
23.13
23.57 23.13
CD8 T cells
CD87.63
T cells 7.86
7.63
7.69
7.86
8.00
7.69
7.69
8.00
7.47
7.69 7.47
CD8 naive
CD83.63
naive 3.74
3.63
3.63
3.74
3.89
3.63
3.78
3.89
3.61
3.78 3.61
CD8 central memoryCD80.71
central memory0.750.71 0.74 0.89 0.77 0.78
0.75 0.74 0.89 0.77 0.78
CD8 effector memory
CD81.69 1.73 1.69
effector memory 1.63 1.62 1.58 1.53
1.73 1.63 1.62 1.58 1.53
CD8 TE CD81.61
TE 1.641.61 1.69 1.60 1.56 1.54
1.64 1.69 1.60 1.56 1.54
CD4 T cells CD414.80
T cells 15.38 14.86 15.67 14.80 14.61
14.80 15.38 14.86 15.67 14.80 14.61
CD4 naive CD44.31
naive 4.344.31 4.08 4.32 4.27 4.00
4.34 4.08 4.32 4.27 4.00
CD4 central memoryCD46.13
central memory6.306.13 6.59
6.30 7.16
6.59 6.62
7.16 6.55
6.62 6.55
CD4 effector memory
CD43.18 3.523.18
effector memory 2.99
3.52 2.95
2.99 2.90
2.95 2.95
2.90 2.95
CD4 terminal effector
CD41.18 1.231.18
terminal effector 1.20
1.23 1.23
1.20 1.01
1.23 1.12
1.01 1.12
γδ T cells γδ T0.68
cells 0.700.68 0.65
0.70 0.70
0.65 0.68
0.70 0.67
0.68 0.67
MAIT/NKT MAIT/NKT
0.41 0.430.41 0.41
0.43 0.44
0.41 0.40
0.44 0.40
0.40 0.40
B cells B cells
5.80 6.005.80 6.116.00 6.19
6.11 6.11
6.19 6.18
6.11 6.18

Helios

Helios
B naive B naive
4.41 4.594.41 4.66
4.59 4.66
4.66 4.63
4.66 4.61
4.63 4.61
B memory B memory
1.35 1.361.35 1.36
1.42 1.42
1.49 1.49
1.45 1.45
1.53 1.53
Plasmablasts Plasmablasts
0.04 0.050.04 0.05
0.03 0.03
0.04 0.04
0.03 0.03
0.04 0.04
Natural killer cells Natural
3.57killer cells 3.613.57 3.61
3.41 3.41
3.67 3.67
3.34 3.34
3.34 3.34
Early natural killer Early natural
1.79 killer 1.801.79 1.80
1.72 1.72
1.86 1.86
1.64 1.64
1.67 1.67
Late natural killerLate1.78
natural killer 1.81 1.78 1.81
1.69 1.69
1.81 1.81
1.70 1.70
1.67 1.67
Monocytes Monocytes
6.41 6.486.41 6.48
6.17 6.17
6.74 6.74
6.46 6.46
6.35 6.35
Classical monocytes 4.81 monocytes4.884.81
Classical 4.88
4.69 4.69
5.15 5.15
4.99 4.99
4.96 4.96
Intermediate mono Intermediate
0.54 mono 0.540.54 0.54
0.50 0.50
0.51 0.51
0.46 0.46
0.43 0.43
Non-classical mono Non-classical
1.07 mono1.061.07 1.06
0.98 0.98
1.08 1.08
1.02 1.02
0.96 0.96
Plasmacytoid
Plasmacytoid dendritic 0.10 dendritic
0.100.10 0.10
0.10 0.10
0.10 0.10
0.09 0.09
0.09 0.09

0.24dendritic 0.250.24
Myeloid dendritic Myeloid 0.25
0.22 0.22
0.28 0.28
0.23 0.23
0.25 0.25
Granulocytes 52.88 51.42 52.86 50.32 52.64 52.90
Granulocytes 52.88 51.42 52.86 50.32 52.64 52.90
Neutrophils 48.71 47.32 48.54 46.69 48.60 48.84
Neutrophils 48.71 47.32 48.54 46.69 48.60 48.84
Basophils 0.59 0.63 0.61 0.65 0.61 0.61
Basophils 0.59 0.63 0.61 0.65 0.61 0.61
Eosinophils 2.82 2.78 2.99 2.21 2.60 2.58
Eosinophils 2.82 2.78 2.99 2.21 2.60 2.58
Treg 0.59 0.65 0.63 0.59 0.56 0.55
Treg 0.59 0.65 0.63 0.59 0.56 0.55
Th1-like 0.95 0.96 0.90 0.92 0.86 0.79
Th1-like 0.95 0.96 0.90 0.92 0.86 0.79
Th2-like 1.02 1.09 0.99 1.08 1.04 1.02
Th2-like 1.02 1.09 0.99 1.08 1.04 1.02
Th17-like 1.32 1.37 1.35 1.45 1.33 1.40
Th17-like 1.32 1.37 1.35 1.45 1.33 1.40

CyTOF
CyTOF XT XT
Table S5. Raw values of population frequencies as percentage of Figure S3. Deming regression of population frequencies between
total live singlet cells of 2 sets of triplicates from Helios and CyTOF XT. 2 sets of triplicates from Table S5. Null hypothesis is accepted at a
The numbers were extracted from Maxpar Pathsetter analysis results. significance level of 0.05, indicating high similarity between the
2 datasets.

77
Beta values of marker intensities of each whole blood sample stained with the Maxpar Direct Immune Profiling Assay from the same donor
on both instruments

Helios Helios Helios CyTOF CyTOF CyTOF

Replicate 1 ReplicateHelios
2 ReplicateHelios
3 ReplicateHelios CyTOF
1 Replicate CyTOF
2 Replicate 3 CyTOF
Replicate 1 Replicate 2 Replicate 3 Replicate 1 Replicate 2 Replicate 3
CD38 4.28 4.26 4.25 4.55 4.32 4.49
CD38 4.28 4.26 4.25 4.55 4.32 4.49
CD14 3.15 3.10 3.14 3.83 3.61 3.80
CD14 3.15 3.10 3.14 3.83 3.61 3.80
CD19 6.06 5.96 6.11 6.91 6.80 6.82
CD19 6.06 5.96 6.11 6.91 6.80 6.82
CD3 7.38 CD37.11 7.38 7.12 7.11 7.00 7.12 7.01 7.00 7.23 7.01 7.23
CD45RA 2.57 2.59
CD45RA 2.572.72 2.59 3.08 2.72 2.78 3.08 3.12 2.78 3.12
CXCR3 0.71 0.71
CXCR3 0.71 0.71 0.71 0.72 0.71 0.72 0.72 0.71 0.72 0.71
CXCR5 7.14 7.18
CXCR5 7.14 7.25 7.18 7.43 7.25 7.41 7.43 7.50 7.41 7.50
CCR4 2.29 2.26
CCR4 2.292.57 2.26 2.65 2.57 2.51 2.65 2.68 2.51 2.68
TCRgd 5.04 4.88
TCRgd 5.044.79 4.88 4.89 4.79 4.95 4.89 4.83 4.95 4.83
CD28 5.07 4.76
CD28 5.074.92 4.76 5.20 4.92 5.29 5.20 5.42 5.29 5.42
CD127 2.58 2.59
CD127 2.582.51 2.59 2.55 2.51 2.50 2.55 2.60 2.50 2.60
CD56 5.00 4.92
CD56 5.004.88 4.92 5.13 4.88 5.16 5.13 5.06 5.16 5.06
CD161 3.68 3.91 3.683.74 3.91 3.71 3.74 3.83 3.71 3.79

Helios
CD161 3.83 3.79

Helios
CD8 6.78 CD86.69 6.786.64 6.69 7.44 6.64 7.33 7.44 7.40 7.33 7.40
CD4 6.91 CD46.69 6.91 6.62 6.69 7.80 6.62 7.30 7.80 7.61 7.30 7.61
CCR7 4.78 CCR7
4.71 4.784.61 4.71 4.41 4.61 4.44 4.41 4.70 4.44 4.70
CD25 2.19 CD25
2.15 2.19 2.25 2.15 2.39 2.25 2.36 2.39 2.35 2.36 2.35
HLADR 4.54 HLADR
4.52 4.544.46 4.52 4.71 4.46 4.78 4.71 4.81 4.78 4.81
CD20 7.11 CD206.89 7.11 6.71 6.89 7.35 6.71 7.32 7.35 7.30 7.32 7.30
IgD 4.85 IgD 4.65 4.854.68 4.65 4.61 4.68 4.67 4.61 4.70 4.67 4.70
CD57 2.65 CD57
2.60 2.652.77 2.60 2.64 2.77 2.56 2.64 2.66 2.56 2.66
CD66b 5.57 CD66b
5.48 5.575.42 5.48 5.73 5.42 5.56 5.73 5.70 5.56 5.70
CD123 4.56 CD123
4.60 4.564.77 4.60 4.45 4.77 5.40 4.45 4.50 5.40 4.50
CD11c 4.20 CD11c
4.92 4.205.08 4.92 4.23 5.08 5.05 4.23 4.63 5.05 4.63

CD27 3.93 CD27

3.98 3.933.70 3.98 3.90 3.70 3.87 3.90 3.94 3.87 3.94

CD45 3.82 CD45

3.81 3.823.80 3.81 4.26 3.80 4.15 4.26 4.30 4.15 4.30
CCR6 4.36 4.32 4.41 4.82 4.70 4.85
CCR6 4.36 4.32 4.41 4.82 4.70 4.85
CD294 4.21 4.14 4.16 4.58 4.44 4.48
CD294 4.21 4.14 4.16 4.58 4.44 4.48
CD16 3.20 3.14 3.10 3.31 3.16 3.34
CD16 3.20 3.14 3.10 3.31 3.16 3.34
CD45RO 3.19 3.13 3.21 3.59 3.53 3.49
CD45RO 3.19 3.13 3.21 3.59 3.53 3.49

CyTOF XT
CyTOF XT
Table S6. Beta (β) values of marker intensities of each replicate from Figure S4. Deming regression of β values of marker intensities from
Helios and CyTOF XT. The numbers were extracted from Maxpar Table S6. Alternative hypothesis is accepted at a significance level
Pathsetter analysis results. of 0.05 because the β values from CyTOF XT are higher than from
Helios, indicating better marker resolution. This translates to a better
separation of negative and positive populations.

78 15
Learn more at fluidigm.com/cytof.xt
Or contact: [email protected]

CORPORATE HEADQUARTERS SALES

2 Tower Place, Suite 2000 North America | +1 650 266 6170 | [email protected]
South San Francisco, CA 94080 USA Europe/EMEA | +33 1 60 92 42 40 | [email protected]
Toll-free: 866 359 4354 in the US and Canada Latin America | +1 650 266 6170 | [email protected]
Fax: 650 871 7152 Japan | +81 3 3662 2150 | [email protected]
standardbiotools.com China (excluding Hong Kong/Macau) | +86 21 3255 8368 | [email protected]
All other Asia-Pacific countries | +1 650 266 6170 | [email protected]

For Research Use Only. Not for use in diagnostic procedures.

Information in this publication is subject to change without notice. Limited Use Label License: The purchase of this Standard BioTools Instrument and/or Consumable
conveys to the purchaser the limited, nontransferable right to use only with Standard BioTools Consumables and/or Instruments respectively except as approved in
writing by Standard BioTools Inc. (f.k.a. Fluidigm Corporation.): www.fluidigm.com/legal/salesterms. Patents: www.fluidigm.com/legal/notices. Trademarks: Standard
BioTools, the Standard BioTools logo, Fluidigm, the Fluidigm logo, the CyTOF XT logo, Cell-ID, CyTOF, CyTOF XT, Direct, EQ, Helios, Immune Profiling Assay, Maxpar and
Pathsetter are trademarks and/or registered trademarks of Standard BioTools Inc. or its affiliates in the United States and/or other countries. All other trademarks are the
sole property of their respective owners. ©2022 Standard BioTools Inc. All rights reserved. 06/2022
FLDM-00462 Rev 02
79
Received: 25 July 2019 Revised: 29 October 2019 Accepted: 5 November 2019
DOI: 10.1002/cyto.b.21858

ORIGINAL ARTICLE

Multi-site reproducibility of a human immunophenotyping

assay in whole blood and peripheral blood mononuclear cells
preparations using CyTOF technology coupled with Maxpar
Pathsetter, an automated data analysis system

Charles Bruce Bagwell1 | Benjamin Hunsberger1 | Beth Hill1 | Donald Herbert1 |

Christopher Bray1 | Thirumahal Selvanantham2 | Stephen Li2 | Jose C. Villasboas3 |
Kevin Pavelko3 | Michael Strausbauch3 | Adeeb Rahman4 | Gregory Kelly4 |
Shahab Asgharzadeh5 | Azucena Gomez-Cabrero5 | Gregory Behbehani6 |
Hsiaochi Chang6 | Justin Lyberger6 | Ruth Montgomery7 | Yujiao Zhao7 |
Margaret Inokuma1 | Ofir Goldberger8 | Greg Stelzer8
1
Verity Software House, Topsham, Maine
2 Abstract
Fluidigm Canada Inc., Markham, Ontario,
Canada High-dimensional mass cytometry data potentially enable a comprehensive charac-
3
Mayo Clinic, Immune Monitoring Core, terization of immune cells. In order to positively affect clinical trials and translational
Rochester, MN
4
clinical research, this advanced technology needs to demonstrate a high reproducibil-
Icahn School of Medicine at Mount Sinai,
New York, New York ity of results across multiple sites for both peripheral blood mononuclear cells
5
Children's Hospital Los Angeles, Los Angeles, (PBMC) and whole blood preparations. A dry 30-marker broad immunophenotyping
California
6
panel and customized automated analysis software were recently engineered and are
Ohio State University, Columbus, Ohio
7 commercially available as the Fluidigm® Maxpar® Direct™ Immune Profiling Assay™.
Yale School of Medicine, New Haven,
Connecticut In this study, seven sites received whole blood and six sites received PBMC samples
8
Fluidigm Corporation, San Francisco, from single donors over a 2-week interval. Each site labeled replicate samples and
California
acquired data on Helios™ instruments using an assay-specific acquisition template.
Correspondence All acquired sample files were then automatically analyzed by Maxpar Pathsetter™
C. Bruce Bagwell, PO Box 247, Topsham, ME
04086. software. A cleanup step eliminated debris, dead cells, aggregates, and normalization
Email: [email protected] beads. The second step automatically enumerated 37 immune cell populations and
performed label intensity assessments on all 30 markers. The inter-site reproducibil-
ity of the 37 quantified cell populations had consistent population frequencies, with
an average %CV of 14.4% for whole blood and 17.7% for PBMC. The dry reagent
coupled with automated data analysis is not only convenient but also provides a high
degree of reproducibility within and among multiple test sites resulting in a compre-
hensive yet practical solution for deep immune phenotyping.

KEYWORDS
cytometry automation, cytometry standardization, kits, percentage precision

© 2019 International Clinical Cytometry Society wileyonlinelibrary.com/journal/cytob Cytometry. 2020;98:146–160.

80
BAGWELL ET AL.

1 | I N T RO D U C T I O N T A B L E 1 Maxpar direct immune profiling assay 30-marker panel

with clones and heavy metals

Multi-site studies have been successfully performed in flow cyto- Target Clone Metal
metry, but only a few multi-site mass cytometry studies have been Anti-human CD45 HI30 89Y
reported (Blazkova et al., 2017; Leipold et al., 2018) and no mass Live/dead 103Rh-Intercalator (500 μM) N/A 103Rh
cytometry-based study has examined the reproducibility of whole Anti-human CD196/CCR6 G034E3 141Pr
blood preparations or dry antibody panels. In mass cytometry, the use
Anti-human CD123 6H6 143Nd
of an inductively coupled plasma mass spectrometer to detect heavy
Anti-human CD19 HIB19 144Nd
metal-tagged probes on a single-cell basis mitigates the issue of spec-
Anti-human CD4 RPA-T4 145Nd
tral overlap between detection channels, easily allowing for the use of
Anti-human CD8a RPA-T8 146Nd
>40 simultaneous measurements.
Anti-human CD11c Bu15 147Sm
Peripheral blood mononuclear cell (PBMC) preparations have use-
Anti-human CD16 3G8 148Nd
ful storage characteristics, which is helpful for doing multi-site studies.
Anti-human CD45RO UCHL1 149Sm
However, immunophenotyping of whole blood specimens is an
industry-standard for clinical trials and other types of clinical studies. Anti-human CD45RA HI100 150Nd

The ability to standardize both PBMC and whole blood immuno- Anti-human CD161 HP-3G10 151Eu

phenotyping worldwide would have far-reaching ramifications. In a Anti-human CD194/CCR4 L291H4 152Sm

typical flow cytometry experiment workflow, several areas of variabil- Anti-human CD25 BC96 153Eu
ity have been identified. Controlling such factors as reagents, sample Anti-human CD27 O323 154Sm
handling, instrument setup, and data analysis can lead to standardiza- Anti-human CD57 HCD57 155Gd
tion (Maecker, McCoy, & Nussenblatt, 2012). Anti-human CD183/CXCR3 G025H7 156Gd
This study is part of an initiative to produce a commercially avail- Anti-human CD185/CXCR5 J252D4 158Gd
able product that addresses many of the factors important in develop- Anti-human CD28 CD28.2 160Gd
ing a standardized immune monitoring assay for mass cytometry. The Anti-human CD38 HB-7 161Dy
system consists of a dry antibody product capable of identifying many
Anti-human CD56/NCAM NCAM16.2 163Dy
important immune populations, an instrument setup template, and
Anti-human TCRgd B1 164Dy
automated cleanup and analysis software that enumerates a broad
Anti-human CD294 BM16 166Er
spectrum of immune cell types. The core of the panel is based on the
Anti-human CD197/CCR7 G043H7 167Er
recommendation of the Human ImmunoPhenotyping Consortium of
Anti-human CD14 63D3 168Er
the Human Immunology Project (Finak et al., 2016; Maecker et al.,
Anti-human CD3 UCHT1 170Er
2012). Eight additional antibodies (CD28, CD45, CD57, CD66b,
Anti-human CD20 2H7 171Yb
CD294, CD161, CXCR5, and TCRγδ) were added to the panel to bet-
ter delineate T-cells, NK cells, and granulocytes, and one marker was Anti-human CD66b G10F5 172Yb

dropped (CD24). In addition to the antibodies, the dry antibody cock- Anti-human HLA-DR LN3 173Yb

tail also includes rhodium for the discrimination of live/dead cells Anti-human IgD IA6-2 174Yb

(Ornatsky et al., 2008). The details of the 30-marker panel are shown Anti-human CD127 A019D5 176Yb
in Table 1, and the workflow is shown in Figure 1.
The analysis of the panel was performed by Maxpar Pathsetter soft-
to the Supporting Information. The purpose of this study is to report in
ware, which uses probability state modeling (PSM) (Bagwell, 2010;
detail on the last stage of validation where the reproducibility of the
Bagwell et al., 2015; Bagwell et al., 2018, Leipold, Maecker, & Stelzer,
kit/analysis system was evaluated by multiple sites for both PBMC and
2016) to obtain frequencies for 37 immune populations (see Table 2 for
whole blood samples from healthy human subjects.
model phenotype definitions) as well as stain assessments for all
30 markers. PSM-derived results have been previously shown to correlate
well with manual gating (Herbert, Miller, & Bagwell, 2012; Li et al., 2018,
2 | MATERIALS AND METHODS
2019; Miller, Hunsberger, & Bagwell, 2012; Wong et al., 2014; Wong,
Hunsberger, Bruce Bagwell, & Davis, 2013). Many different validation
2.1 | Study sites
tests needed to be performed prior to releasing this product. These tests
included liquid versus dry panel, intra-assay repeatability, intermediate A total of seven sites (six in the United States plus Fluidigm Canada)
precision, manual gating versus modeling correlations, and inter-site were selected to participate in these reproducibility studies. These
reproducibility. Most of these validations are presented in a publicly avail- sites are designated as Sites 1, 2, 3, 4, 5, 6, and 7. Site 1 received
able white paper. Deep Immune Profiling with the Maxpar Direct Immune whole blood products in Week 1 of the study, for which it is desig-
Profiling System 400247 A2) and data from other tests have been added nated as Site 1A, and in the second week of the study received whole

81
BAGWELL ET AL.

F I G U R E 1 Assay workflow. Based on the broad immune cell phenotyping flow panels for the Human Immune Project (Maecker et al., 2012),
the Maxpar Direct Immune Profiling Assay was designed as an optimized panel of 30 dry antibodies plus DNA intercalators in a single tube for
staining whole blood and PBMC. Data were acquired on a Fluidigm Helios and analyzed using Maxpar Pathsetter, a customized automated
analysis system powered by GemStone 2.0. Pathsetter software automatically cleans the data file by eliminating dead cells, debris, aggregates,
and normalization beads. Modeling software then identifies and enumerates a broad spectrum of immune populations and presents the results in
summary reports [Color figure can be viewed at wileyonlinelibrary.com]

blood products from a second draw from the same donor, for which it 2.4 | PBMC specimens
is designated as Site 1B. Site 1 did not participate in the PBMC part of
One lot of cryopreserved PBMC from a single healthy donor was
the study. Sites 2, 3, and 4 received whole blood and PBMC samples in
obtained from a commercial biological specimen supply source
Week 1, and Sites 5, 6, and 7 received the products in the second
(Discovery Life Sciences) and reserved as the reference lot for the
week. All sites were given careful instructions on the staining and anal-
study. Two vials of cryopreserved PBMC were shipped on dry ice to
ysis procedures, and Fluidigm Field Application Specialists were on
each of six sites. The PBMC samples were thawed based on the man-
hand to provide general guidance on all the procedures.
ufacturer's (Discovery Life Sciences) recommendations, which was to
thaw in serum-free media with no anti-aggregate.
2.2 | Whole blood collection and shipping
Human whole blood was obtained from Discovery Life Sciences 2.5 | PBMC staining
(Huntsville, AL). Whole blood from a single healthy donor was col-
A vial of cryopreserved PBMC was thawed and washed. The viability
lected into eight individual BD Vacutainer® blood collection tubes
and cell count were determined and the cells were washed in CSB.
containing heparin as an anticoagulant. Two tubes of the whole blood
After the wash, the cells were resuspended in CSB to a concentration
were shipped on cold packs to each study site overnight in a
of 6 × 107 cells/ml. FC receptors were blocked by adding 5 μl of
temperature-controlled shipping container.
Human TruStain FcX to 3 × 106 cells in 50 μl and incubated for
10 min. About 215 μl of CSB was then added to the PBMC. About
2.3 | Whole blood staining 270 μl of the PBMC was added directly added to each of the four dry
antibody tubes for antibody staining (see Table 1). After a 30-min
An additional heparin blocking step was performed (100 U/ml) for
incubation, the cells were washed twice in CSB, followed by fixation
20 min at room temperature to reduce nonspecific binding between
in 1.6% paraformaldehyde for 10 min. Following fixation, the cells
metal-tagged antibodies and eosinophils (Rahman, Tordesillas, & were spun to a pellet, the fixative was removed, and the pellet was
Berin, 2016). Afterward, 270 μl of blood was added directly to four resuspended in 1 ml of the 125 nM Cell-ID Intercalator-Ir and incu-
dry antibody tubes and allowed to incubate for 30 min at room tem- bated overnight at 4 .
perature. Immediately following staining, erythrocytes were lysed by
the addition of 250 μl of Cal-Lyse directly to the staining tube. The
tubes were gently vortexed and allowed to incubate for 10 min at
2.6 | Sample acquisition
room temperature followed by the addition of 3 ml of Maxpar water Following the overnight incubation, the PBMC fixed cells were
and an additional 10 min of incubation. The tubes were washed three washed twice in CSB and twice with Maxpar Cell Acquisition Solution
times in Maxpar Cell Staining Buffer (CSB) followed by fixation in (CAS) with a final resuspension of the cells at 1 × 106 cells/ml in CAS
1.6% paraformaldehyde for 10 min. Following fixation, the cells were containing 0.1× EQ™ Four Element Calibration Beads. Whole blood
spun to a pellet, the fixative removed, and the pellet was resuspended sample acquisition was also performed the next day post staining on a
in 1 ml of the 125 nm Cell-ID™ Intercalator-Ir (Ornatsky et al., 2008) Helios system utilizing CyTOF® Software version 6.7.1016 using the
and incubated overnight at 4 (See Figure 1 for the assay workflow). Maxpar Direct Immune Profiling Assay template. All instruments were

82
BAGWELL ET AL.

TABLE 2 Immune cell populations and model definitions

Index Populations Model phenotypes

1 Lymphocytes CD3 T cells + B cells + NK cells + plasmablasts
2 CD3 T cells CD8 T cells + CD4 T cells + γδ T cells + MAIT/NKT cells
3 CD8 T cells CD3+ CD66b- CD19- CD8+ CD4- CD14- CD161- TCRgd- CD123- CD11c-
4 CD8 naïve CD8 T cells + CD45RA+ CCR7+ CD27+
5 CD8 central memory CD8 T cells + CD45RA- CCR7+ CD27+
6 CD8 effector memory CD8 T cells + CCR7- CD27+
7 CD8 terminal effector CD8 T cells + CCR7- CD27-
8 CD4 T cells CD66b- CD3+ CD8- CD4+ CD14- TCRgd- CD11c-
9 CD4 naïve CD4 T cells + CD45RA+ CCR7+ CD27+
10 CD4 central memory CD4 T cells + CD45RA- CCR7+ CD27+
11 CD4 effector memory CD4 T cells + CD45RA- CCR7- CD27+
12 CD4 terminal effector CD4 T cells + CD45RA- CCR7- CD27-
13 Tregs CD4 T cells + CD25+ CD127- CCR4+
14 Th1-like CD4 T cells + CXCR3+ CCR6- CXCR5- CCR4-
15 Th2-like CD4 T cells + CXCR3- CCR6- CXCR5- CCR4+
16 Th17-like CD4 T cells + CXCR3- CCR6+ CXCR5- CCR4+
17 γ T cells CD66b- CD3+ CD8dim,- CD4- CD14- TCRgd dim,+
18 MAIT/NKT cells CD66b- CD3+ CD4- CD14- CD161+ TCRgd- CD28+ CD16-
19 B cells CD3- CD14- CD56- CD16 dim,- CD19+ CD20+ HLA-DR dim,+
20 B naïve B cells + CD27-
21 B memory B cells + CD27+
22 Plasmablasts CD3- CD14- CD16-,dim CD66b- CD20- CD19+ CD56- CD38++ CD27+
23 NK cells CD14- CD3- CD123- CD66b- CD45RA+ CD56 dim,+
24 NK early NK cells + CD57-
25 NK late NK cells + CD57+
26 Monocytes CD3- CD19- CD56- CD66b- HLA-DR+ CD11c+
27 Monocytes classical Monocytes + CD14+ CD38+
28 Monocytes transitional Monocytes + CD14 dim CD38 dim
29 Monocytes non-classical Monocytes + CD14- CD38-
30 DCs pDCs + mDCs
31 pDCs CD3- CD19- CD14- CD20- CD66b- HLA-DR dim,+ CD11c- CD123+
32 mDCs CD3- CD19- CD14- CD20- HLA-DR dim,+ CD11c dim,+ CD123- CD16 dim,- CD38
dim,+ CD294- HLA-D
33 Granulocytes Neutrophils + basophils + eosinophils + CD66b- neutrophils
34 Neutrophils CD66b dim,+ CD16+ HLA-DR-
35 Basophils HLA-DR- CD66b- CD123 dim,+ CD38+ CD294+
36 Eosinophils CD14- CD3- CD19- HLA-DR- CD294+ CD66b dim,+
37 CD66b- neutrophils CD3- CD19- CD66b- CD56- HLA-DR- CD123- CD45-

The above table shows the 37 immune cell populations enumerated and their associated model phenotypes.
The modeling algorithm is designed to fit the measurements in the order listed by the phenotype. Nomenclature such as TCRγδ dim,+ means that dim to
positive events were selected. Occasionally the same marker is modeled twice, where the first time is a broader classification and the last time is a more
specific classification. See Section 4 for details on the subsetting and staging rationales for monocytes, CD8 T-cells, and CD4 T-cells.

equipped with a WB Injector, and all samples were acquired in CAS tuning and bead sensitivity test, the system was preconditioned with
containing 0.1× EQ beads. Prior to the start of the study, all instru- CAS. A minimum of 400,000 events for whole blood and 300,000
ments were evaluated to ensure performance at above the minimum events for PBMC were acquired per file at a typical acquisition rate of
Helios system specifications for calibration. Following the instrument 250–500 events/s.

83
BAGWELL ET AL.

F I G U R E 2 Cleanup and analysis

Cen-se0 maps: The top two panels are
Cen-se0 maps created from the QC
measurements: DNA1, DNA2, Live/
Dead, Beads, Event Length, Residual,
Center, Width, and Offset. The top-
left panel represents the raw
normalized data from one file and the
top-right the associated cleaned
exported data. In the top-left panel, A
(dark gray) are the live intact events, B
(blue) are the low-DNA1 or debris
events, C (yellow) are the
normalization beads, D (blue) are
events with zero pulse-processing
parameters (Residual, Center, Width,
and Offset), E (red) are “not cleaned
events” with high Residual and Event
Lengths, F (red) are true aggregates
with high DNA1 and DNA2
intensities, G (yellow) are bead/cell
aggregates, and H (red) are coincident
ion clouds with low and high center
values. The top-right panel is the Cen-
se0 map of only the “cleaned” events.
The bottom panel shows the same
data with all markers selected after
cleanup and modeling [Color figure
can be viewed at
wileyonlinelibrary.com]

2.7 | Data normalization automated analysis of a second model, which also uses PSM to iden-
tify and label the major immune cell populations in sample files.
After acquisition, data were normalized using the CyTOF Software
This system is integrated with dimensionality-reduction mapping
v. 6.7.1016. This method normalizes the data to a global standard,
known as Cauchy Enhanced Nearest-neighbor Stochastic Embedding
called a bead passport, determined for each log of EQ beads. This
(Cen-se0 ™), which generates a visual display of high-dimensional data
passport contains a profile of mean Di counts of all the masses for a
labeled with the major cell populations. Figure 2 shows a Cen-se0 map of
particular lot of the beads as determined by multiple measurements
only QC measurements from one of the whole blood files in the study
during the manufacture of the EQ beads. The normalization factor is before and after the cleanup procedure (see top-left and right panels) as
the ratio of passport median Di values to bead singlet population well as a map of all markers after full analysis (see bottom-right panel).
median Di values of the encoding isotopes. Isotopes in the EQ beads All analyses were done on the same mid-level PC (Intel® Core™ i7-6700
cover the mass range measurable on the CyTOF instrument. The nor- CP @3.40 GHz RAM: 24 GB x64-based processor). The average run time
malization factors for mass channels between the encoding isotopes for the whole blood Cleanup Stage was 37.3 s with a range of
are linearly interpolated. All mass channel values for all events are 36.5–37.9. The run time statistics for the other parts of the study were
then multiplied by these normalization factors to obtain the normal- PBMC Cleanup Stage: 33.2 s (32.2–39.7), whole blood Phenotyping and
ized values, and data are written to the normalized file. Cen-se' Stage: 207.7 s (137.3–227.9), PBMC Phenotyping and Cen-se'
Stage: 233.6 s (212.8–282.1). The complete average analysis time for
the whole blood samples was 4.1 min and for PBMCs, 4.4 min.
2.8 | Data analysis
FCS files generated by the Helios were analyzed by Maxpar
Pathsetter, an automated analysis system powered by GemStone™
3 | RESULTS
2.0.41 (Verity Software House, Topsham, ME). Initial analyses process
3.1 | Whole blood
raw normalized FCS3.0 files with a specially designed Cleanup PSM
model. The Cleanup model leverages Gaussian pulse-processing A total of 32 whole blood-derived files from seven different sites
parameters such as Center, Width, Offset, and Residual as well as were analyzed by the cleanup phase of the analysis (see Table 3 for a
DNA intercalators to eliminate unwanted events. Subsequent to summary of the results). On average, 70.9% of the events were con-
cleanup, the program produces new FCS3.0 files consisting of only sidered desirable “live intact cells”; 26.9% were excluded because they
intact live singlet cells. This new cleaned file is then processed by an were classified as dead cells, debris, true aggregates, aborted pulses,

84
TABLE 3 Whole blood cleanup summary statistics

Multi-site whole blood reproducibility study: Cleanup statisticsa

Site Replicate %Cleanb %Excluded %Beads %Unclassc %Debris %Dead %Aggs CeO2 ratio Acq rate %CD19+ CD3+ d %CD14+ CD3+ e Total cells Run time f
BAGWELL ET AL.

Site 1A 1 65.1 32.7 1.9 0.4 9.9 0.1 13.8 1.6 406.1 0.3 0.5 400,000 37.1
2 61.4 36.1 1.6 1.0 14.6 0.2 12.3 1.7 410.7 0.3 1.0 400,000 37.7
3 65.8 32.4 1.4 0.5 11.9 0.1 11.7 1.3 353.0 0.2 0.4 400,000 37.9
4* 51.9 45.0 1.2 1.8 27.3 0.8 9.3 1.5 444.9 0.2 21.5 400,000 37.5
Site 2 1 72.4 24.9 2.5 0.2 7.4 0.1 8.4 1.8 244.1 0.2 0.5 400,000 37.5
2 72.5 24.8 2.5 0.2 6.4 0.1 9.1 1.8 269.4 0.2 0.5 398,479 37.2
3 71.0 26.8 2.1 0.2 9.5 0.1 8.2 1.8 272.9 0.2 0.5 400,000 37.7
4 75.0 22.5 2.4 0.1 4.9 0.1 9.0 1.8 246.3 0.2 0.6 400,000 37.0
Site 3 1 69.2 26.9 3.6 0.3 12.8 0.0 7.5 0.8 264.7 0.2 0.3 400,000 37.2
2 61.4 34.8 3.3 0.5 20.6 0.0 7.7 1.9 260.4 0.2 0.5 400,000 37.2
3 67.9 29.3 2.3 0.5 16.9 0.1 5.8 1.4 184.4 0.2 0.7 398,553 37.4
4 72.3 23.0 4.5 0.2 6.7 0.0 9.8 2.0 349.0 0.2 0.4 400,000 37.1
Site 4 1 74.7 23.4 1.7 0.1 4.0 0.1 11.6 2.9 330.1 0.3 0.3 398,439 37.0
2 72.3 26.2 1.0 0.5 5.2 0.1 13.6 2.8 389.9 0.3 0.4 400,000 37.4
3 74.1 24.8 0.7 0.4 3.4 0.1 14.5 2.8 384.2 0.3 0.3 400,000 37.1
4 70.2 29.0 0.6 0.2 4.4 0.1 16.5 2.9 402.4 0.4 0.5 400,000 37.4
Site 1B 1 69.3 28.5 1.9 0.2 9.2 0.1 11.5 1.6 327.3 0.2 0.5 400,000 37.3
2 68.6 29.7 1.4 0.3 10.2 0.1 11.3 1.6 333.6 0.2 0.5 400,000 37.3
3* 43.4 53.8 1.2 1.6 38.8 1.3 8.1 2.0 496.9 0.2 21.2 400,000 37.7
4 74.5 24.3 0.9 0.3 7.0 0.1 11.1 1.8 305.6 0.2 0.8 400,000 37.2
Site 5 1* 67.1 29.0 3.3 0.5 15.5 0.3 8.0 1.1 293.5 0.2 18.7 400,000 37.7
2 71.5 26.0 2.3 0.2 4.2 0.1 13.7 1.2 386.5 0.2 3.3 400,000 37.3
3 72.5 24.8 2.5 0.2 3.3 0.1 14.5 1.1 373.8 0.2 1.6 400,000 37.3
4 71.0 27.0 1.8 0.1 4.7 0.1 14.0 1.2 391.4 0.2 0.8 400,000 37.4
Site 6 1 79.0 20.1 0.8 0.2 3.0 0.0 11.8 1.4 278.4 0.2 1.0 400,000 37.0
2 76.7 22.6 0.5 0.2 5.8 0.0 10.8 1.4 286.0 0.1 1.0 392,922 36.5
3 78.5 20.7 0.7 0.1 4.3 0.0 9.8 1.4 264.9 0.1 1.2 400,000 37.2
4 77.7 21.6 0.5 0.1 4.3 0.1 11.6 1.4 290.4 0.1 1.0 395,841 37.2
Site 7 1 78.1 19.4 2.4 0.1 3.7 0.2 11.3 1.6 317.7 0.2 0.3 400,000 37.3
2 81.5 16.1 2.3 0.1 3.0 0.2 9.4 1.6 260.1 0.1 0.3 400,000 37.2
3 80.7 17.6 1.6 0.1 2.4 0.2 12.2 1.7 333.6 0.2 0.5 400,000 37.2
(Continues)

85

BAGWELL ET AL.

or coincident ion clouds; 1.8% were normalization beads; and 0.4%

the normalization bead percentage. %Unclass are the percentage of unclassified events. The %Clean+%Excluded+%Bead+%Unclass fields add to 100%. %Debris are the sub DNA1 events that are not beads, %
The %Clean column quantifies the percentage of total events that were exported without debris, dead cells, aggregates, and normalization beads. %Excluded are the non-bead excluded events and %Beads are
Run time f
were unclassified. Approximately 13.2% of the excluded events were

Aggs are the high DNA1 events, and %Dead are the events that are Live/Dead+. CeO+ ratio is an indicator of plasma temperature and should be less than 3.0. The acquisition rate is the number of acquired
events/s, which is recommended to be approximately 350 events/s. The %CD19 + CD3+ and %CD14 + CD3+ columns are indicators of cell aggregation and coincident ion clouds. Total Cells are the total
debris, 10.9% were high DNA1 aggregates, and the %dead cell count

37.2
37.3
was low at 0.2%. All files had CeO+ ratios, a measure of plasma tem-
perature, of less than 3. The average acquisition rate was approxi-
Total cells
400,000
399,507
mately 326.8 events/s, and the average % of aggregates was
reasonably low (%CD19 + CD3+ and %CD14 + CD3+ less than 0.2
and 2.6%, respectively). A total of 400,000 events were considered by

number of events acquired for analysis, and Run Time is the length of time in seconds for cleanup analysis. Three files were excluded due to background signal in the Er168 channel.
%CD14+ CD3+ e

the cleanup routine, and the average time spent in this step was
approximately 37 s.
The deep immunophenotyping frequency results for whole blood
are summarized in Table 4. The left side of the table shows the enu-
0.4
2.6

merated populations, and the numbers indicate the percentages of

%CD19+ CD3+ d

live intact cells in each of the replicates from all seven sites. Three
replicates (Site 1A Rep 4, Site 1B Rep 3, and Site 5 Rep 1) were
excluded due to background signal in the Er168 channel (see
Section 4 for details). Figure 3 summarizes the inter-site reproducibil-
0.1
0.2

ity of all populations with both SDs and %CV of each population.
Means, SDs, and %CVs from Sites 1A, 2, 3, and 4 were calculated sep-
Acq rate
304.6
326.8

arately from Sites 1B, 5, 6, and 7 because they were from a different
sample. Statistics from both sets of sites were averaged. The percent-
ages of live intact cells for each population and SDs are summarized in
CeO2 ratio

the top panel, and the %CVs are presented as a bar graph in the bot-
tom panel. The inter-site average %CV was 14.4%, ranging from 2.3
1.7
1.7

to 96.6%, with higher %CVs generally associated with very low-

frequency populations. The intra-site reproducibility is summarized in
%Aggs
10.7
10.9

Table 5 and had an average %CV of 7.9%.

%Dead

3.2 | PBMC
0.2
0.2

A total of 24 FCS3.0 PBMC-derived files from six different sites were

%Debris

analyzed by the cleanup phase of the analysis (See Table 6 for a sum-
2.5
9.0

mary of the results). On average, 76.7% of the events were consid-

ered desirable “live intact cells”; 21.4% were excluded because they
%Unclassc

were classified as dead cells, debris, true aggregates, aborted pulses,

or coincident ion clouds; 1.8% were normalization beads, and 0.08%
0.1
0.4
Multi-site whole blood reproducibility study: Cleanup statisticsa

were unclassified. Approximately 4.2% of the excluded events were

All FCS files were automatically analyzed by the Cleanup model.

debris, 0.9% were dead, and 11.4% were high DNA1 aggregates. All
%Beads

files had CeO+ ratios of less than 3.5. The average acquisition rate
1.2
1.8

was approximately 300 events/s, and the average % of double-

positive aggregates was reasonably low (%CD19 + CD3+ and %
%Excluded

CD14 + CD3+ less than 0.3 and 2.3%, respectively). Approximately

15.7
26.9

300,000 events were considered by the cleanup routine, and the

average time spent in this step was approximately 33 s.
%Cleanb

The deep immunophenotyping results are summarized in Table 7.

83.0
70.9

The left side of the table shows the enumerated populations, and the
(Continued)

numbers indicate the percentages of live intact cells from the four
Replicate

replicates from all six sites. Figure 4 summarizes the inter-site repro-
ducibility of the percentages with both SDs as well as %CV. The per-
4

centiles and SDs are summarized in the top panel, and the %CVs
Averages
TABLE 3

presented as a bar graph in the bottom panel. The percentages, SDs,

and %CVs were an average of Cohort 1 (Week 1) and 2 statistics. The
Site

average and median %CV were 17.7 and 13.7%, respectively. The

86
TABLE 4 Multi-site whole blood reproducibility study

Whole blood reproducibility study: Population percentagesa

Donor week 1b Same donor week 2c

BAGWELL ET AL.

Blood source
Site 1Ad Site 2 Site 3 Site 4 Site 1B Site 2 Site 6 Site 7
Site
e
Replicate 1 2 3 1 2 3 4 1 2 3 4 1 2 3 4 1 2 4 2 3 4 1 2 3 4 1 2 3 4

Lymphocytes 28.2 27.8 29.4 25.7 27.3 26.2 28.2 24.9 26.7 24.7 25.5 27.1 26.9 25.6 27.0 22.1 22.1 21.2 18.3 19.4 18.0 21.5 20.3 20.3 19.3 22.4 22.4 20.8 22.3

CD3 T cells 19.9 19.4 20.5 18.0 19.0 18.3 19.6 17.7 18.6 17.4 18.4 19.2 18.9 17.6 19.1 15.7 15.5 14.8 12.8 13.9 12.7 15.2 14.3 14.4 13.7 15.9 15.9 15.1 15.8

CD8 T cells 4.4 4.3 4.6 4.1 4.3 4.1 4.5 4.1 4.3 4.0 4.2 4.5 4.4 4.2 4.4 3.3 3.3 3.1 2.8 3.0 2.8 3.3 3.0 3.1 2.9 3.5 3.5 3.3 3.5

CD8 naïve 2.7 2.6 2.8 2.6 2.6 2.6 2.8 1.9 2.3 2.0 2.4 2.7 2.6 2.5 2.6 2.0 2.0 1.9 1.7 1.8 1.7 2.0 1.8 1.8 1.7 2.0 2.1 2.0 2.1

CD8 central memory 0.6 0.5 0.6 0.4 0.4 0.4 0.4 0.9 0.8 0.7 0.5 0.6 0.5 0.5 0.5 0.4 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.4 0.3 0.3 0.3

CD8 effector memory 1.0 1.0 1.1 1.0 1.1 1.0 1.1 1.1 1.1 1.1 1.1 1.0 1.1 1.1 1.1 0.8 0.8 0.8 0.7 0.7 0.8 0.9 0.8 0.9 0.8 1.0 0.9 0.9 1.0

CD8 terminal effector 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.1 0.2 0.2 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.2 0.2 0.1 0.1

CD4 T cells 12.8 12.5 13.1 11.4 12.1 11.7 12.5 11.3 11.7 10.9 11.6 12.3 12.1 11.7 12.3 10.1 9.8 9.5 8.0 8.8 7.9 9.7 9.1 9.2 8.9 9.9 10.0 9.5 10.0

CD4 naïve 2.8 2.7 2.9 3.0 3.3 3.0 3.4 1.5 1.9 1.8 2.2 2.7 2.8 2.8 2.8 2.3 2.2 2.2 2.2 2.4 2.1 2.4 2.2 2.2 2.2 2.3 2.4 2.2 2.4

CD4 central memory 4.3 4.3 4.2 3.3 3.5 3.4 3.6 4.4 4.4 3.9 3.9 3.9 3.5 2.8 4.0 3.2 3.2 3.0 1.8 2.1 2.0 2.9 2.7 2.7 2.5 2.7 2.7 2.5 2.9

CD4 effector memory 4.5 4.4 4.8 4.0 4.2 4.1 4.4 4.3 4.2 4.2 4.4 4.5 4.7 5.1 4.3 3.6 3.5 3.4 3.2 3.4 2.9 3.4 3.2 3.4 3.2 3.8 3.9 3.8 3.7

CD4 terminal effector 1.1 1.1 1.2 1.1 1.2 1.1 1.1 1.0 1.2 1.1 1.1 1.2 1.1 1.0 1.2 1.0 0.9 1.0 0.8 0.9 0.9 1.0 0.9 0.9 0.9 1.1 1.0 1.0 1.0

γδ T cells 1.5 1.4 1.5 1.4 1.4 1.4 1.5 1.3 1.4 1.4 1.5 1.4 1.4 1.5 1.5 1.2 1.2 1.2 1.1 1.1 1.1 1.2 1.2 1.1 1.1 1.3 1.3 1.2 1.2

MAIT/NKT cells 1.2 1.2 1.2 1.1 1.1 1.1 1.1 1.0 1.1 1.1 1.1 1.0 1.0 0.2 0.9 1.2 1.1 1.0 1.0 1.0 1.0 1.0 0.9 1.0 0.9 1.1 1.2 1.1 1.1

B cells 3.6 3.5 3.8 2.9 3.2 3.0 3.4 2.2 3.0 2.2 1.6 3.0 2.9 3.0 2.8 2.4 2.5 2.3 1.7 2.0 1.7 2.0 1.9 1.8 1.7 2.3 2.2 2.0 2.2

B Naïve 3.2 3.1 3.4 2.6 2.8 2.7 3.1 1.9 2.6 1.9 1.4 2.7 2.5 2.6 2.5 2.1 2.2 2.1 1.5 1.8 1.5 1.7 1.7 1.5 1.5 2.0 2.0 1.8 2.0

B memory 0.3 0.4 0.4 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.2 0.3 0.3 0.3 0.3 0.2 0.3 0.2 0.2 0.3 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2

Plasmablasts 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

NK cells 4.7 4.9 5.1 4.8 5.1 4.9 5.2 5.0 5.2 5.0 5.4 4.9 5.1 5.0 5.1 4.1 4.1 4.0 3.7 3.5 3.6 4.3 4.1 4.1 3.9 4.3 4.2 3.7 4.2

NK early 3.9 4.0 4.3 3.7 4.0 3.8 4.1 4.4 4.5 4.4 4.7 3.9 4.0 4.0 4.0 3.4 3.4 3.3 2.8 2.7 2.7 3.5 3.4 3.3 3.1 3.3 3.2 2.8 3.2

NK late 0.8 0.8 0.9 1.0 1.1 1.1 1.1 0.6 0.6 0.7 0.7 1.1 1.1 1.0 1.1 0.7 0.7 0.7 0.9 0.8 0.9 0.8 0.8 0.8 0.8 1.0 1.0 0.9 1.0

Monocytes 3.6 3.7 3.8 4.0 4.0 3.8 3.9 2.7 3.0 2.8 2.6 3.2 2.6 1.9 2.5 3.0 3.1 3.3 2.6 4.4 4.0 4.2 4.4 4.6 4.4 4.6 4.8 4.8 4.7

Monocytes classical 3.2 3.3 3.4 3.5 3.6 3.4 3.5 2.0 2.4 2.1 2.2 2.8 2.1 1.5 2.1 2.7 2.9 3.0 2.3 4.1 3.7 3.8 4.0 4.2 4.1 4.3 4.4 4.5 4.4

Monocytes Intermediate 0.2 0.2 0.3 0.2 0.2 0.2 0.2 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.1 0.2 0.2 0.1 0.1 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2

Monocytes nonclassical 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.3 0.3 0.3 0.2 0.2 0.2 0.2 0.2 0.1 0.1 0.1 0.1 0.2 0.1 0.2 0.2 0.2 0.2 0.2 0.2 0.1 0.1

DCs 0.3 0.5 0.3 0.2 0.2 0.2 0.3 0.3 0.4 0.5 0.3 0.2 0.3 0.2 0.3 0.2 0.2 0.2 0.3 0.3 0.3 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2

pDCs 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1

mDCs 0.2 0.4 0.2 0.2 0.2 0.1 0.2 0.2 0.1 0.4 0.2 0.1 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.1 0.1 0.2 0.2 0.2

Granulocytes 64.1 64.3 63.1 65.3 63.6 65.8 63.4 65.4 64.4 66.9 66.3 65.3 66.4 66.4 66.6 71.2 70.8 70.9 74.2 72.7 74.6 70.8 71.0 71.5 73.0 69.0 68.7 70.2 69.1

Neutrophils 61.7 62.1 61.0 63.3 61.8 63.8 61.6 62.7 60.9 63.8 63.9 61.2 63.1 59.3 61.9 69.4 68.9 68.7 72.3 70.5 72.8 69.0 69.2 69.9 71.0 67.4 67.1 68.6 67.5

(Continues)

87

BAGWELL ET AL.

0.6

1.0

0.0

0.3

0.7

0.5
4

0.6

1.0

0.0

0.3

0.7

0.6

0.6
3

0.7

0.9

0.0

0.3

0.7

0.6
2
Site 7

0.7

0.9

0.0

0.3

0.7

0.5
1

0.6

1.0

0.1

0.3

0.6

0.5

0.5
4

0.6

1.0

0.1

0.3

0.7

0.6

0.5
3

0.6

1.0

0.2

0.3

1.2

0.9

0.2
2
Site 6

0.5

1.0

0.3

0.7

0.6

0.5
1

0.6

1.0

0.6

0.3

0.7

0.5

0.4
4

0.6

0.9

0.7

0.3

0.7

0.6

0.4
3
Site 2

0.5

0.9

0.5

0.3

0.6

0.5

0.4
2
Same donor week 2c

0.6

0.9

0.7

0.3

0.7

0.6

0.5
4

F I G U R E 3 Whole blood reproducibility. The top panel shows the

0.7

0.9

0.4

0.3

0.7

0.5

mean and ± SD percentage of live intact cells for all 37 evaluated

2
Site 1B

populations across all seven sites. The bottom panel shows the
0.6

0.9

0.4

0.7

0.6

associated %CVs for each population where the average was 14.4%
1

[Color figure can be viewed at wileyonlinelibrary.com]

0.6

1.0

3.2

0.3

1.1

0.8

0.6

Four replicates per site were processed except for three files with high background in 163Er channel (see Section 4).
The above table summarizes all the frequency results for all sites and populations in terms of percent live intact cells.
4

0.6

1.0

5.7

0.3

1.0

0.8

0.6

intra-site reproducibility is summarized in Table 8 and had an average

and median %CV of 8.4 and 4.5%, respectively, for all sites and
0.7

1.0

1.6

0.3

1.1

0.8

0.6
2

populations.
Site 4

0.6

1.0

2.6

0.3

1.0

0.8

0.6
1

4 | DISCUSSION
0.7

1.0

0.7

0.3

1.1

0.9

0.5
4

0.7

0.9

1.5

0.2

1.2

1.0

0.3

The average event inclusion percentage for this study was approxi-
3

mately 70% for whole blood (see Table 3) and 76.7% for PBMC (see
0.7

0.9

1.9

0.2

1.3

1.1

0.4
2

Table 6), which are generally comparable to gate-based inclusion per-

Site 3

0.7

0.9

1.1

0.2

1.3

1.0

0.3

centages (data not shown). The site-to-site variability is probably

All samples were stained with Maxpar Direct Immune Profiling Assay.

Same donor was drawn in week two and sent to right four test sites.

either due to different environmental factors during the shipping of

0.7

1.0

0.1

0.4

0.9

0.8

0.6
4

the samples or to slightly different site specimen handling techniques.

0.7

1.1

0.2

0.4

0.8

0.7

0.6

The increase in %CD14+ CD3+ in Table 3 is due to the inclusion of

the three files with a high CD14 Er168 background.

Same donor of whole blood sent to left four test sites.
Whole blood reproducibility study: Population percentagesa

0.7

1.0

0.1

0.4

0.9

0.7

0.6
2

The average acquisition rate for both the whole blood and PBMC
Site 2

0.7

1.1

0.4

0.7

0.6

studies was approximately 300 events/s. Although the acquisition

Population percentages are of intact live cells.
1

system can be set for faster rates, the Poisson nature of ion cloud for-
0.8

0.8

0.6

0.4

1.0

0.9

0.7
3
Donor week 1b

mation creates more coincident clouds at faster rates. Most of these

coincident events are removed in the cleanup stage, but the routine is
0.7

1.0

0.5

0.4

1.0

0.8

0.6
2
Site 1Ad

not 100% effective in eliminating these events. A rate between

0.7

1.1

0.6

0.4

0.9

0.6
1e
(Continued)

250 and 350 events/s is currently recommended by Fluidigm

(Fluidigm, 2018).
The Cleanup model exports FCS3.0 data that are not only avail-
CD66b- Neuts

able for PSM automated analysis but also for other types of cytometry
Blood source

Eosinophils
TABLE 4

analysis as well. For investigators interested in oncological samples,

Th17-like
Basophils
Replicate

Th1-like

Th2-like
Tregs

the DNA selection parts of the Cleanup model can easily be

Site

deactivated in order that DNA hyperdiploid populations are not

e
a

88
BAGWELL ET AL.

TABLE 5 Whole blood intra-site reproducibility

Whole blood intra-site reproducibility %CV

Population Site 1A Site 2 Site 3 Site 4 Average Wk 1 Site 1B Site 5 Site 6 Site 7 Average Wk 2
Lymphocytes 2.8 4.1 4.0 1.5 3.1 2.4 8.3 5.2 4.2 4.0
CD3 T cells 2.6 3.7 3.1 2.2 2.9 2.9 8.4 6.0 2.8 3.9
CD8 T cells 3.3 3.5 3.7 3.1 3.4 3.6 6.5 7.9 2.6 4.2
CD8 naïve 3.5 4.0 7.8 3.8 4.8 4.3 8.4 7.0 2.6 5.1
CDS central memory 4.9 5.0 18.5 8.4 9.2 6.0 5.3 12.4 2.8 8.0
CDS effector memory 2.5 4.5 1.9 3.7 3.1 2.7 4.0 9.9 4.4 4.1
CDS terminal effector 3.7 4.1 6.8 9.7 6.1 12.0 9.7 6.4 8.7 7.5
CD4 T cells 2.6 4.0 3.0 1.9 2.9 2.9 10.3 4.7 2.9 3.9
CD4 naïve 4.1 5.7 17.7 2.1 7.4 2.0 7.1 2.9 3.8 5.9
CD4 central memory 3.3 4.5 7.4 15.3 7.6 8.0 23.3 3.9 7.5 9.0
CD4 effector memory 4.2 4.8 0.6 8.6 4.6 2.2 6.0 6.3 1.7 4.3
CD4 terminal effector 9.4 6.1 4.2 7.3 6.7 3.6 8.1 10.6 1.4 6.4
γδ T cells 3.9 3.8 4.7 3.1 3.9 1.5 5.9 7.3 4.3 4.3
MAIT/NKT cells 3.4 1.9 2.4 10.9 4.6 4.0 2.6 12.0 4.1 5.1
B cells 4.0 6.5 23.5 1.9 9.0 3.1 11.8 12.5 5.7 8.7
B naïve 4.4 6.6 25.4 2.0 9.6 2.6 10.7 13.6 5.9 9.0
B memory 2.8 6.9 12.8 3.9 6.6 7.0 20.6 8.5 5.0 8.2
Plasmablasts 11.3 9.7 15.5 21.7 14.5 10.3 16.3 15.8 11.7 14.1
NK cells 4.3 4.1 8.5 1.7 4.6 1.6 9.0 3.2 9.4 5.2
NK early 4.3 5.0 7.0 1.9 4.6 2.1 12.6 5.8 10.1 5.9
NK late 4.5 1.5 19.6 4.0 7.4 0.9 6.9 8.8 7.5 6.8
Monocytes 2.7 1.9 4.2 10.7 4.9 5.4 21.2 2.0 1.1 6.0
Monocytes classical 2.7 2.0 3.6 13.5 5.5 5.6 22.2 2.8 1.2 6.6
Monocytes transitional 4.4 9.3 29.5 7.9 12.8 6.8 10.6 16.5 2.7 11.2
Monocytes non-classical 5.2 5.8 13.1 6.6 7.7 2.4 19.5 7.5 8.7 8.5
DCs 35.0 7.9 21.4 16.7 20.3 5.4 8.5 4.7 4.7 13.8
pDCs 5.8 3.0 9.7 1.6 5.0 11.0 15.0 9.6 7.0 7.5
mDCs 43.4 11.9 26.3 24.1 26.4 3.2 6.1 6.1 3.7 16.8
Granulocytes 1.2 1.9 1.6 0.8 1.4 0.2 2.6 2.1 1.2 1.4
Neutrophils 1.0 l.8 2.1 2.6 1.9 0.5 2.5 2.1 1.2 1.7
Basophils 5.6 2.6 1.5 7.9 4.4 4.3 6.5 7.0 1.9 4.6
Eosinophils 14.4 3.8 3.5 4.1 6.5 2.0 5.5 3.1 2.6 5.1
CD66b- Neuts 8.6 73.7 38.9 53.2 43.6 36.6 35.9 66.1 37.4 43.8
Tregs 5.0 2.6 25.1 3.7 9.1 5.6 7.0 4.5 4.6 7.5
Th1-like 2.9 7.0 6.4 4.9 5.3 5.9 8.6 30.0 6.2 8.6
Th2-like 3.9 4.0 5.2 5.2 4.6 7.0 11.0 19.3 7.6 7.5
Th17-like 4.9 5.6 14.5 1.9 6.7 3.8 13.2 29.9 3.9 9.4
Mean 6.4 6.6 10.9 7.7 7.9 5.2 10.7 10.4 5.5 7.9

removed. However, if these measurements are deactivated, the num- The staging approach for CD8 T-cells (see Table 2) was to first
ber of true aggregates in the exported “cleaned” file is likely to model the downregulation of CCR7 and CD27 to stratify events into
increase. The data obtained in the multi-site study were generated three compartments: naïve + central memory, effector memory, and
using a prototype panel lot. Three out of 24 runs were excluded from terminal effector. CD45RA was found not to be a good modeling
the data presented due to background signals in the Er168 channel, marker for staging because of its relatively wide line-spread (data not
which has been eliminated in subsequent manufacturing lots. shown) and branched nature (Inokuma, Maino, & Bagwell, 2013). The

89
BAGWELL ET AL.

TABLE 6 PBMC cleanup summary statistics

Multi-site PBMC reproducibility study: Cleanup statisticsa

% % % % % % % CeO2 Acq %CD19+ %CD14+ Total Run

Sites Replicates Clean b Excluded Beads Unclass c Debris Dead Aggs ratio rate CD3+ d CD3+ e Cells timef
Site 2 1 76.9 21.7 1.4 0.1 4.0 0.3 11.4 1.8 316.5 0.3 1.2 300,000 32.2
2 75.9 22.8 1.3 0.1 5.2 0.3 11.1 1.9 298.8 0.3 1.4 300,000 33.2
3 77.3 21.5 1.1 0.1 4.1 0.2 11.1 1.9 313.2 0.3 1.1 300,000 33.0
4 78.2 20.5 1.3 0.1 4.3 0.2 10.1 1.2 214.2 0.2 1.1 288,975 32.4
Site 3 1 73.7 24.0 2.2 0.1 6.8 0.1 11.4 2.1 266.1 0.4 1.3 298,335 33.1
2 68.8 28.4 2.6 0.1 8.7 0.1 13.2 2.1 303.5 0.5 1.6 400,000 39.9
3 76.6 16.7 6.6 0.1 3.0 0.1 7.8 2.0 248.3 0.2 1.2 300,000 33.6
4 75.2 19.4 5.3 0.1 4.0 0.1 9.0 2.0 279.3 0.3 1.9 300,000 32.9
Site 4 1 72.1 27.0 0.8 0.1 6.2 0.6 15.0 2.8 413.2 0.6 1.5 300,000 33.7
2 66.4 32.6 0.7 0.2 6.2 0.4 18.2 3.3 414.9 0.5 1.6 300,000 33.3
3 79.2 19.5 1.2 0.1 4.0 0.3 10.6 3.0 300.0 0.4 1.4 300,000 32.5
4 76.7 21.7 1.5 0.1 5.1 0.2 11.3 2.9 295.6 0.5 1.5 300,000 32.8
Site 5 1 75.7 22.8 1.4 0.1 2.2 0.2 15.5 1.1 403.8 0.5 1.4 300,000 32.9
2 78.2 20.0 1.7 0.1 2.0 0.0 13.5 1.1 361.0 0.4 1.3 300,000 32.6
3 75.9 22.5 1.5 0.1 2.8 0.2 15.2 1.2 394.7 0.4 7.9 300,000 32.7
4 77.4 20.2 2.2 0.2 8.2 0.4 7.6 1.2 241.5 0.2 14.2 300,000 32.5
Site 6g 1 75.6 23.3 1.0 0.1 4.6 3.1 11.0 1.4 254.2 0.2 1.2 300,000 33.0
2 74.4 24.9 0.6 0.1 6.1 3.6 10.2 1.4 244.7 0.3 1.9 297,593 32.6
3 83.6 15.3 1.1 0.0 2.5 2.8 5.4 1.4 127.6 0.1 2.0 300,000 32.8
4 75.7 19.5 4.7 0.1 4.9 4.0 6.5 1.4 164.5 0.2 2.3 300,000 32.4
Site 7 1 81.7 17.4 0.8 0.0 1.9 1.3 11.7 1.7 317.1 0.3 1.3 300,000 32.9
2 80.7 18.6 0.7 0.0 1.7 1.6 12.9 1.6 356.7 0.3 2.0 300,000 33.0
3 82.0 17.4 0.6 0.0 1.4 1.0 12.4 1.7 339.4 0.4 1.8 295,265 33.0
4 83.5 15.9 0.6 0.0 1.4 0.9 11.1 1.9 308.3 0.3 1.7 300,000 32.7
Mean 76.7 21.4 1.8 0.08 4.2 0.9 11.4 1.8 299.1 0.3 2.3 303,340 33.2
a
All samples were stained with Maxpar Direct Immune Profiling Assay.
b
Percent of total events. %Cleaned+%Excluded+%Beads+%Unclassified = 100.
c
Percent of events that were not classified into the cell types Cleaned, Excluded, or Beads.
d
Percent of CD19+ CD3+ double positives of CD19+ singlets + CD3+ singlets.
e
Percent of CD14+ CD3+ double positives of CD14+ singlets + CD3+ singlets.
f
Units of seconds.
g
The first acquisition of Site 6 samples had insufficient EQ Beads for normalization. Samples were spun down and resuspended again in fresh CAS/0.1 EQ
Beads to acquire data for analysis.

system then used a combinatory analysis system called TriCOM to traditional CD14 and CD16 (Picozza, Battistini, & Borsellino, 2013).
divide the first stage into its naïve and central memory components. The patterns produced by CD14 and CD38 were found to classify
The staging approach for CD4 T-cells (see Table 2) was to model the analogous subpopulations while improving the overall reproducibility
downregulation of CD45RA, CCR7, and CD27 to create the four of the results (data not shown).
stages: naïve, central memory, effector memory, and terminal effector. The data presented in Tables 4 and 7 summarize all the cell popu-
The CD4 T-cell terminal effector was assumed to be CD45RA− lation frequency results obtained from the whole blood and PBMC
because CD45RA+ events were generally not observed in any sample studies. An inspection of these tables shows the high degree of repro-
in this study (see Figure 5) and it has been recognized that there are a ducibility of the system for almost all immune populations. Figures 3
few if any CCR7− CD45RA+ events in the CD4 T-cell compartment and 4 summarize the inter-site variability of the whole blood and
for healthy individuals (Seder & Almed, 2003). PBMC studies. The populations are ordered from the highest percent-
Subclassification of monocytes into Classical, Transitional, and age (left) to the lowest (right) in order to better appreciate the
Non-classical used CD14 and CD38 (see Table 2) instead of the more general effect of counting error increasing the magnitude of CVs for

90
TABLE 7 Multi-site PBMC reproducibility study

Multi-site PBMC reproducibility study: Population percentagesa

Site 2 b Site 3 Site 4 Site 5 Site 6 Site 7

BAGWELL ET AL.

Site
Replicate 1c 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4
Lymphocytes 67.1 65.9 66.4 65.7 67.0 67.5 67.4 67.2 65.6 65.7 66.0 66.0 75.6 73.6 65.5 62.4 64.3 64.3 70.1 62.7 68.1 68.3 69.4 69.8
CD3 T cells 50.5 49.8 50.1 50.1 48.6 49.6 48.0 48.4 48.0 47.7 46.1 44.9 50.8 51.6 46.2 40.1 48.6 48.5 46.4 43.5 48.2 48.6 50.6 50.3
CD8 T cells 10.3 10.1 10.2 10.1 10.3 10.4 10.2 10.2 10.3 10.2 10.4 10.3 11.4 11.2 9.8 9.8 10.6 10.6 12.6 10.4 10.6 10.2 10.5 10.5
CD8 naïve 6.2 6.1 6.2 6.1 6.0 6.0 5.9 5.9 6.2 6.4 6.3 6.2 6.2 6.4 5.9 5.3 6.4 6.4 7.4 6.0 6.2 5.8 6.1 5.9
CD8 central memory 0.4 0.4 0.5 0.5 0.5 0.6 0.6 0.6 0.6 0.4 0.5 0.5 0.5 0.4 0.4 0.2 0.5 0.5 0.4 0.4 0.3 0.4 0.4 0.4
CD8 effector memory 2.0 2.1 2.0 2.0 2.2 2.2 2.0 2.1 1.7 1.5 1.7 1.6 2.6 2.5 2.0 2.3 1.9 1.8 2.1 1.9 2.3 2.3 2.2 2.4
CD8 terminal effector 1.6 1.4 1.5 1.6 1.7 1.6 1.7 1.6 1.7 1.8 1.9 2.0 2.2 1.9 1.5 2.0 1.9 1.9 2.7 2.0 1.9 1.8 1.8 1.9
CD4 T cells 32.4 32.0 32.1 32.3 29.9 30.8 29.5 30.0 31.5 31.1 29.2 28.4 29.8 30.7 28.6 23.9 29.7 29.8 24.5 25.4 30.4 30.3 31.6 31.2
CD4 naïve 12.7 12.5 12.4 12.6 11.1 11.4 11.0 11.3 13.5 13.8 12.3 12.6 12.8 13.3 12.7 10.2 11.4 12.2 11.3 9.9 12.4 12.6 13.0 12.6
CD4 central memory 6.4 6.3 7.0 6.5 6.6 6.7 6.7 6.9 6.7 6.1 5.9 5.5 4.4 4.9 4.7 3.3 6.8 6.4 4.1 5.6 5.6 5.1 5.5 5.3
CD4 effector memory 9.8 9.8 9.2 9.7 8.4 8.4 7.9 7.8 6.5 6.6 6.6 6.0 9.0 8.7 8.1 7.2 7.3 7.1 5.3 6.0 8.3 8.6 9.1 9.1
CD4 terminal effector 3.5 3.5 3.6 3.5 3.8 4.3 3.9 4.1 4.8 4.6 4.4 4.2 3.7 3.8 3.1 3.3 4.1 4.2 3.8 3.9 4.1 3.9 4.0 4.1
γδ T cells 4.3 4.4 4.3 4.2 4.6 4.7 4.5 4.6 4.4 4.4 4.5 4.5 6.0 5.3 4.4 4.5 4.4 4.5 5.4 4.6 4.8 4.5 4.6 4.7
MAIT/NKT cells 3.5 3.3 3.5 3.5 3.7 3.7 3.8 3.6 1.8 1.9 1.9 1.8 3.6 4.4 3.3 1.8 4.0 3.6 4.0 3.1 2.4 3.6 3.9 4.0
B cells 7.1 6.6 6.9 6.4 7.6 7.6 8.0 8.0 7.9 8.1 8.6 9.1 8.5 8.0 7.4 6.2 5.3 5.2 5.1 5.8 7.4 7.3 7.3 7.3
B naïve 5.9 5.4 5.6 5.3 6.4 6.4 6.8 6.7 6.7 6.7 7.3 7.8 7.3 6.9 6.3 5.3 4.3 4.3 4.3 4.8 6.3 6.2 6.1 6.2
B memory 1.1 1.1 1.1 1.0 1.1 1.1 1.1 1.2 1.2 1.3 1.3 1.3 1.1 1.0 1.0 0.7 0.9 0.8 0.7 0.9 1.0 1.0 1.0 1.0
Plasmablasts 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1
NK cells 9.6 9.5 9.4 9.1 10.8 10.3 11.4 10.9 9.6 10.0 11.3 11.9 16.4 14.1 11.9 16.2 10.3 10.6 18.6 13.4 12.5 12.3 11.6 12.2
NK early 2.7 2.7 2.7 2.7 3.5 3.4 3.8 3.5 2.9 3.0 3.3 3.6 4.5 3.9 3.1 4.2 3.2 3.3 6.0 4.3 3.6 3.6 3.3 3.5
NK late 6.8 6.8 6.7 6.4 7.3 6.9 7.6 7.3 6.7 7.0 8.0 8.4 11.9 10.2 8.8 12.0 7.1 7.3 12.6 9.1 8.9 8.7 8.3 8.7
Monocytes 24.2 24.6 24.9 24.0 24.3 22.8 25.2 24.9 25.3 25.9 26.0 25.9 15.7 17.3 18.0 14.0 25.5 25.0 21.4 26.0 22.1 22.1 21.1 21.1
Monocytes classical 19.4 19.4 19.9 19.0 19.6 18.2 20.5 20.1 19.0 19.4 19.4 19.2 11.5 13.4 14.6 11.1 19.4 18.9 14.8 19.7 17.8 17.8 17.0 16.9
Monocytes transitional 2.9 3.1 2.9 2.9 2.7 2.6 2.8 2.9 3.3 3.0 3.3 3.3 2.1 2.3 2.8 2.5 3.5 3.4 3.0 3.4 2.6 2.6 2.5 2.5
Monocytes non-classical 1.9 2.1 2.0 2.1 1.9 1.9 1.9 1.9 3.1 3.4 3.3 3.3 2.1 1.7 0.7 0.4 2.6 2.6 3.6 2.9 1.7 1.7 1.6 1.7
DCs 0.9 0.9 0.9 0.8 1.0 1.0 0.9 0.8 1.0 1.2 1.0 1.0 0.7 0.7 0.3 0.2 1.3 1.3 1.5 1.2 1.1 1.0 0.9 1.0
pDCs 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.1 0.0 0.2 0.2 0.3 0.2 0.2 0.2 0.2 0.2
mDCs 0.7 0.7 0.7 0.7 0.9 0.9 0.7 0.6 0.8 1.0 0.8 0.7 0.6 0.6 0.2 0.1 1.1 1.1 1.3 1.0 0.9 0.8 0.8 0.8
Tregs 0.8 0.8 0.8 0.8 0.7 0.8 0.7 0.8 0.5 0.5 0.5 0.5 0.6 0.6 0.6 0.4 0.4 0.1 0.4 0.5 0.7 0.7 0.7 0.7
Th1-like 2.0 1.8 2.0 1.9 1.4 1.3 1.4 1.4 1.1 1.1 1.2 1.4 1.9 1.8 1.6 1.0 0.4 0.0 0.3 0.3 0.3 0.3 0.3 0.2
(Continues)

91

BAGWELL ET AL.

4.1
3.1
4
5.0
2.1
3
5.1
1.6
2
Site 7

5.2
1.6
1
4.5
0.4
4
3.0
0.7
3
0.0
5.7
2
Site 6

5.6
0.3
1
3.3
1.5
4
3.7
1.6

F I G U R E 4 Peripheral blood mononuclear cells (PBMC)

reproducibility. The top panel shows the mean and ±SD percentage of
4.0
1.8

live intact cells for all 37 evaluated populations. Absent from this plot
2

are the granulocyte, neutrophils, basophils, eosinophils, and CD66b−

Site 5

3.8
1.7

granulocytes. The bottom panel shows the associated %CVs for each
1

population, where the average was 17.7%. The percentages, SDs, and
The above table summarizes all the frequency results for all sites and populations in terms of percent live intact cells.

%CVs were an average of Cohort 1 (Week 1) and 2 statistics [Color

3.0
1.5
4

figure can be viewed at wileyonlinelibrary.com]

2.8
2.1
3

low-frequency populations. The average %CV for all 37 populations

2.9
2.4

was 14.4% for whole blood and 17.7% for PBMC. The slight increase
2
Site 4

in variability for the PBMC may be due in part to the extra cell manip-
3.1
2.5
1

ulations for this type of preparation. A high %CV was observed for
the population labeled as CD66b− neutrophils in whole blood mainly
3.0
2.9
4

due to its low frequency.

The upper panel insets with the ±SD ranges show a high degree of
2.8
2.8
3

reproducibility even among many of the very low-frequency

populations. Some populations are better defined by the panel than
3.2
3.0

All samples were stained with Maxpar Direct Immune Profiling Assay.
2
Multi-site PBMC reproducibility study: Population percentagesa

others, which explain some of the variability in the %CVs for

Site 3

2.9
2.9

populations with similar frequencies, and additional markers may be

included to enhance identification in studies focused on low-

2.2
2.7

frequency cell populations. The PBMC portion of this study is reason-

ably comparable to the multi-site study published by Leipold

2.1
2.6

et al. (2018).
Population percentages are of live intact cells.
3

Tables 5 and 8 summarize the intra-site reproducibility of the

Four replicates per site were processed.
2.2
2.7

whole blood and PBMC studies. As expected, the average and median
2
Site 2 b

intra-site %CV's are lower than the inter-site %CV's due to slight site-
2.1
2.6
c
1

to-site biases. Some of the high intra-site %CV's for both whole blood
(Continued)

and PBMC were due to outliers in the relatively small number of repli-
cates. There was some disparity in intra-site %CV's across all
populations among the sites in the study, which was more pro-
nounced for low-frequency cell types.
Th17-like
Replicate
TABLE 7

The dry nature of the reagent in this assay eliminates most

Th2-like

pipetting errors and reduces overall preparation time. An important

Site

feature of this system is that additional reagents can be added to

b
a

92
BAGWELL ET AL.

TABLE 8 PBMC intra-site reproducibility

Intra-site PBMC reproducibility %CV

Population Site 2 Site 3 Site 4 Site 5 Site 6 Site 7 Average Median

Lymphocytes 1.0 0.3 0.3 9.1 5.0 1.2 2.8 1.1
CD3 T cells 0.5 1.4 3.1 11.2 5.1 2.4 4.0 2.7
CD8 T cells 0.9 1.3 0.9 8.2 9.2 1.6 3.7 1.4
CD8 naïve 1.0 0.7 1.6 8.0 9.0 2.6 3.8 2.1
CD8 central memory 10.0 4.6 13.5 25.8 13.4 18.3 14.3 13.4
CD8 effector memory 3.2 4.8 5.0 11.4 5.0 2.8 5.4 4.9
CD8 terminal effector 5.4 1.7 5.6 13.7 19.2 2.4 8.0 5.5
CD4 T cells 0.5 1.9 5.0 10.6 10.3 2.0 5.1 3.5
CD4 naïve 1.0 1.5 5.4 11.2 8.6 2.0 5.0 3.7
CD4 central memory 4.7 1.6 8.0 16.8 20.6 3.9 9.3 6.4
CD4 effector memory 3.2 3.9 4.4 9.6 14.8 4.1 6.7 4.3
CD4 terminal effector 1.2 5.7 5.3 9.6 5.1 2.1 4.8 5.2
γδ T cells 2.0 1.2 0.9 14.6 9.9 2.5 5.2 2.3
MAIT/NKT cells 2.1 2.3 3.1 32.8 11.3 20.9 12.1 7.2
B cells 4.3 2.9 6.4 13.2 5.7 0.6 5.5 5.0
B naïve 4.7 3.1 7.4 13.3 5.6 1.2 5.9 5.1
B memory 2.8 3.9 4.2 18.0 13.6 3.0 7.6 4.1
Plasmablasts 3.7 12.4 13.2 27.0 13.2 5.7 12.5 12.8
NK cells 2.3 4.0 10.1 14.4 29.0 3.3 10.5 7.0
NK early 0.9 4.6 9.4 14.8 30.4 3.8 10.7 7.0
NK late 3.0 3.8 10.4 14.3 28.4 3.1 10.5 7.1
Monocytes 1.6 4.4 1.2 11.0 8.5 2.5 4.9 3.5
Monocytes classical 1.8 5.0 1.1 12.9 12.6 2.6 6.0 3.8
Monocytes transitional 2.6 4.5 4.1 12.0 6.8 1.8 5.3 4.3
Monocytes non-classical 3.8 1.2 4.3 65.3 15.6 2.6 15.5 4.0
DCs 4.3 10.9 10.3 61.5 9.5 7.2 17.3 9.9
pDCs 0.2 5.2 5.6 60.7 13.1 4.7 14.9 5.4
mDCs 5.4 14.3 13.1 61.9 9.4 8.7 18.8 11.3
Tregs 2.7 3.3 8.0 13.5 58.6 4.9 15.2 6.5
Th1-like 5.3 5.6 12.1 24.8 66.7 18.6 22.2 15.3
Th2-like 2.4 4.5 3.9 8.2 73.8 10.6 17.2 6.4
Th17-like 1.9 2.8 20.4 8.1 146.3 33.8 35.6 14.3
Mean 2.8 4.0 6.5 20.2 21.7 5.9 10.2 6.2
Median 2.5 3.8 5.4 13.4 11.9 2.9 6.7 4.6

The 37 tested populations appear in the first column, and the %CVs of the four replicate PBMC samples are summarized for each site. The means and
medians of the %CVs for all populations and sites appear on the outside rows and columns.

evaluate new populations because there are numerous open heavy 2008) that can create maps of hundreds of thousands of events in
metal channels. The Maxpar Pathsetter software is also designed for 1 min or less.
users to easily amend the models to take advantage of new markers The dry nature of the reagent coupled with automated data analy-
and cell types. sis is not only convenient but also provides a high degree of reproduc-
The performance of the analysis system was designed to do a full ibility within and among multiple test sites, whether they are
and automated analysis in less than 5 min. The Cen-se mapping sys-
0 analyzing whole blood or PBMC samples. This new mass cytometry
tem is a high-resolution and highly parallelized variant of the t-SNE assay provides a comprehensive yet practical solution for deep
algorithm (van der Maaten, 2009, 2014; van der Maaten & Hinton, immune phenotyping.

93
BAGWELL ET AL.

ACKNOWLEDGMENTS Leipold, M. D., Obermoser, G., Fenwick, C., Kleinstuber, K., Rashidi, N.,
McNevin, J. P., … Maecker, H. T. (2018). Comparison of CyTOF assays
The authors wish to acknowledge the helpful editing from Dmitry across sites: Results of a six-center pilot study. Journal of Immunologi-
Bandura, Tony Shuga, and Dorothy McDuffie. This work could not have cal Methods, 453, 37–43.
Li, S., Majonis, D., Bagwell, C. B., Hunsberger, B. C., Baranov, V. I., &
been done without the resources of both Verity Software and Fluidigm.
Ornatsky, O. (2018). An efficient human whole blood workflow using
CyTOF technology: A lyophilized 30-plex antibody panel coupled with
automated data analysis. Journal of Immunology, 200, 120.2.
CONF LICT OF IN TE RE ST
Li, S., Majonis, D., Bagwell, C. B., Hunsberger, B. C., Baranov, V. I., &
Authors Bagwell, Hunsberger, Hill, Herbert, Bray, Selvanantham, Li, Ornatsky, O. (2019). A robust human immunophenotyping workflow
using CyTOF technology coupled with Maxpar Pathsetter, an auto-
Inokuma, Goldberger, and Stelzer are currently employed or were
mated data analysis software. Journal of Immunology, 202.
employed by either Verity Software House or Fluidigm Corporation. Maecker, H. T., McCoy, J. P., & Nussenblatt, R. (2012). Standardizing
Author Inokuma is a consultant for both Verity Software House and immunophenotyping for the human immunology project. Nature
Fluidigm Corporation. This manuscript describes a component of the Reviews Immunology, 12, 191–200.
Miller, D. T., Hunsberger, B. C., & Bagwell, C. B. (2012). Automated analy-
product, Fluidigm Maxpar Pathsetter, which was a collaborative effort
sis of GPI-deficient leukocyte flow cytometric data using GemStone.
between these two companies. Cytometry Part B, Clinical Cytometry, 82, 319–324.
Ornatsky, O. I., Lou, X., Nitz, M., Schafer, S., Sheldrick, W. S., Baranov, V. I.,
… Tanner, S. D. (2008). Study of cell antigens and intracellular DNA by
DIS CLOSURES identification of element-containing labels and metallointercalators
using inductively coupled plasma mass spectrometry. Analytical Chem-
Fluidigm, Cell-ID, CyTOF, EQ, Helios, Maxpar, and Pathsetter are istry, 80, 2539–2547.
trademarks and/or registered trademarks of Fluidigm Corporation in Picozza, M., Battistini, L., & Borsellino, G. (2013). Mononuclear phagocytes
the United States and/or other countries. All other trademarks are the and marker modulation: When CD16 disappears, CD38 takes the
stage. Blood, 122, 456–457.
sole property of their respective owners. For research use only, not
Rahman, A. H., Tordesillas, L., & Berin, M. C. (2016). Heparin reduces non-
for use in diagnostic procedures. specific eosinophil staining artifacts in mass cytometry experiments.
Cytometry Part A, 89, 601–607.
Seder, R. A., & Almed, R. (2003). Similarities and differences in CD4+ and CD8
RE FE R ENC E S + effector and memory T cell generation. Nature Immunology, 4, 835–842.
van der Maaten, L. (2009). Learning a Parametric Embedding by Preserving
Bagwell, C. (2010). Probability state modeling: A new paradigm for
Local Structure. In AISTATS. TiCC. Tilburg, The Netherlands: Tilburg
cytometric analysis. In V. Litwin & P. Marder (Eds.), Flow cytometry in
University.
drug discovery and development (p. 281). Hoboken, NJ: John Wiley and
van der Maaten, L. (2014). Accelerating t-SNE using tree-based algorithms.
Sons Inc.
Journal of Machine Learning Research, 15, 1–21.
Bagwell, C. B. (2018). Chapter 2: High-dimensional modeling for cyto-
van der Maaten, L., & Hinton, G. (2008). Visualizing data using t-SNE. Jour-
metry: Building rock solid models using GemStonetm™ and Verity
nal of Machine Learning Research, 9, 2579–2605.
Cen-se'™ high-dimensional t-SNE mapping. In T. S. Hawley &
Wong, L., Hill, B. L., Hunsberger, B. C., Bagwell, C. B., Curtis, A. D., &
R. G. Hawley (Eds.), Methods in molecular biology (Vol. 1678, p. 2018).
Davis, B. H. (2014). Automated analysis of flow cytometric data for
New York: Springer Science+Business Media LLC.
measuring neutrophil CD64 expression using a multi-instrument com-
Bagwell, C. B., Hunsberger, B. C., Herbert, D. J., Munson, M. E., Hill, B. L.,
patible probability state model. Cytometry Part B, Clinical Cytometry,
Bray, C. M., & Preffer, F. I. (2015). Probability state modeling theory.
88(4), 227–235.
Cytometry Part A, 87, 646–660.
Wong, L., Hunsberger, B. C., Bruce Bagwell, C., & Davis, B. H. (2013).
Bagwell, C. B., Leipold, M., Maecker, H., & Stelzer, G. (2016). High-
Automated quantitation of fetomaternal hemorrhage by flow cyto-
dimensional modeling of peripheral blood mononuclear cells from a Helios
metry for HbF-containing fetal red blood cells using probability state
Instrument. Seattle, Washington: Washington State Convention
modeling. International Journal of Laboratory Hematology, 35, 548–554.
Center.
Blazkova, J., Gupta, S., Liu, Y., Gaudilliere, B., Ganio, E. A., Bolen, C. R., …
Furman, D. (2017). Multicenter systems analysis of human blood SUPPORTING INF ORMATION
reveals immature neutrophils in males and during pregnancy. Journal
of Immunology, 198, 2479–2488. Additional supporting information may be found online in the
Finak, G., Langweiler, M., Jaimes, M., Malek, M., Taghiyar, J., Korin, Y., … Supporting Information section at the end of this article.
JP, M. C. (2016). Standardizing flow cytometry immunophenotyping
analysis from the human ImmunoPhenotyping consortium. Scientific
Reports, 6, 20686.
Fluidigm (2018). Maxpar Human Immune Monitoring Panel Kit Cell
Staining and Data Acquisition. In: Fluidigm, editor. PN PRD027 C1
Protocol.
Herbert, D., Miller, D., & Bagwell, C. (2012). Automated analysis of flow
cytometric data for CD34+ stem cell enumeration using a probability
state model. Cytometry Part B, Clinical Cytometry, 82B, 313–318.
Inokuma, M. S., Maino, V. C., & Bagwell, C. B. (2013). Probability state
modeling of memory CD8(+) T-cell differentiation. Journal of Immuno-
logical Methods, 397, 8–17.

94
Mass Cytometry: a robust platform for the comprehensive
immunomonitoring of CAR-T-cell therapies

By reprograming T-cell specificity, chimaeric antigen recep-

Development and robustness of a mass
tors’ (CARs) function and persistence have changed the land-
cytometry customized panel for CAR and non-
scape of cancer immunotherapy with remarkable efficacy
CAR immune cell determination
against a range of B-cell malignancies and have many
prospective applications.1,2 Investigating the immune mecha- The Maxpar Direct Immune Profiling System, bringing
nisms that underlie the success of CAR-T-cell therapy is a together mass cytometry technology, a 30-antibody panel and
crucial challenge.3,4,5 The advent of time-of-flight mass fully automated reporting, is a high-dimensional immune
cytometry (CyTOF) has enabled high-dimensional and unbi- profiling assay designed to broadly look at the phenotypes
ased examination of complex systems.6,7 We present here the and functions of immune cell subsets through a single-tube
first contribution of mass cytometry to the comprehensive view. So far, no metal-tagged antibodies for mass cytometry
immunomonitoring of CAR-T-cell therapies. that identify CAR-T cells have been described. Therefore,
We constructed an original platform to determine whether we customized the Maxpar panel by developing a cadmium
identification of CAR-T cells and immune cells lacking CAR (106/116Cd)–anti-biotin CD19Fc (106/116 cadmium metal iso-
(non-CAR) and their functional states in peripheral blood topes, Fluidigm, Canada; pure anti-biotin antibody and
would be possible in a single-pass mass cytometry assay. We CD19Fc Detection Reagent, biotin, Milteny Biotec, Bergisch
therefore customized a commercially available immune panel Gladbach, Germany) that identify CAR-T cells and validated
(Maxpar Direct Immune Profiling System, Fluidigm, San its robustness by comparison with routine assays. CAR-T
Francisco, CA, USA) for mass cytometry with a CAR-T-cell cells were usually assessed by flow cytometry, using the CAR
detection reagent along with complementary markers.8 Detection Reagents (Miltenyi), and by quantitative poly-
The mass cytometry assay was developed in patients with merase chain reaction (qPCR) using an assay commercialized
refractory diffuse large B-cell lymphoma (DLBCL) and acute to quantify HIV-1 (Generic HIV DNA Cell� test for research
lymphoblastic leukemia (ALL) treated in our CAR-T-cell use, Biocentric, Bandol, France).10,11 These two assays per-
programme by tisagenlecleucel (Kymriah, Novartis, Bale, formed similarly (r = 06782, P < 00001 by Spearman corre-
Suisse; Table S1). Patients received a lymphodepletive regi- lation; n = 44; Fig 1A and Figure S1). Over a broad range of
men, CAR-T-cell infusion and post CAR-T-cell care accord- frequencies from 012 to 22% CD3+ CAR-T cells, we showed
ing to standard practices and all gave written informed that, for the detection of CAR-T cells, results obtained with
consent.9 our 106/116Cd–anti-biotin CD19Fc used for mass cytometry

ª 2021 British Society for Haematology and John Wiley & Sons Ltd
British Journal of Haematology, 2021, 194, 779–792

95
Correspondence

Fig 1. Robustness of mass cytometry for the monitoring of blood chimaeric antigen receptor (CAR)-T cells and immune cells. Peripheral blood
mononuclear cells from 10 patients having received CAR-T cells were stained and analyzed to determine the populations of CAR-T and immune
cells using mass cytometry, flow cytometry and quantitative polymerase chain reaction (qPCR). For flow cytometry, data acquisition and analyses
were performed on a Navios flow cytometer (Beckman Coulter, Villepinte, France); for qPCR, data were acquired on a Light Cycler� 480 (Roche
Diagnostics, Meylan, France); for mass cytometry, data were acquired on a Helios machine and analyzed using Maxpar Pathsetter software
(Gemstone, Verity Software House, Topsham, ME, San Francisco, CA, USA with Fluidigm) with integrated dimensionality reduction mapping
(Cen-seTM). (A) Routine assays for peripheral CAR-T-cell determination using flow cytometry and qPCR. Representative biaxial plot of CD3+ and
CD19Fc+ CAR-T cells (left panel). Flow cytometry for measurement of the percentage of CD19Fc+ CAR-T cells (Miltenyi Biotec) and qPCR for
quantification of long-terminal repeat (LTR) sequences (Generic HIV DNA Cell�test for research use, Biocentric, Bandol, France) performed sim-
ilarly (n = 44; P < 0001 using a Spearman correlation [GraphPad software, Prism, San Diego, CA, USA]; middle panel). Representative quantifi-
cation of CAR-T cells over time using the two assays. qPCR copies of HIV; percentage of CD3+/CD19Fc+ CAR-T cells using flow cytometry;
absolute value (/mm3) of CD3+/CD19Fc+ CAR-T cells (right panel). (B) Strong agreement between the percentages of CD19Fc+ CAR-T cells as
determined by mass and flow cytometry (n = 72). Pearson correlation between the quantification of CAR-T cells using our cadmium (106/119Cd)–
anti-biotin, CD19Fc and the CD19 CAR Detection reagent (Miltenyi) for respectively mass and flow cytometry; the linear regression line is shown
on the left panel. Representative biaxial plot of percentage of CD3+/CD19Fc+ CAR-T cells using mass and flow cytometry and Bland–Altman vali-
dation method (right panel). (C) Validation of the customized mass cytometry panel for T-cell composition (n = 52). Mann–Whitney U test
(GraphPad; left panel) together with the Bland–Altman validation method (middle panel) confirmed that mass cytometry correlated strongly with
routine flow cytometry assays for measurement of the main subsets of T cells (CD3, 4, 8, -HLA-DR, 69, 25, 45RA, 45RO). These data
also allowed to convert CAR–T cells detected by mass or flow cytometry into the absolute value of viable CD3+ cells, as measured by single-
platform flow cytometry (right panel). CMF, flow cytometry; CMM, mass cytometry. [Colour figure can be viewed at wileyonlinelibrary.com]

strongly correlated with those obtained by flow cytometry We also completed the panel with four other anti-human
(Pearson correlation; r = 095, P < 005; n = 72; confirmed markers (NKP30, PD1, CD163 and CD69, tagged with 159Tb,
175
by the validation method of Bland–Altman; Fig 1B). Lu, 165Ho and 162Dy, respectively; Fluidigm) and checked that

ª 2021 British Society for Haematology and John Wiley & Sons Ltd
British Journal of Haematology, 2021, 194, 779–792

96
Correspondence

Fig 2. Automated identification of cell populations and their functional states by mass cytometry. (A) t-Distributed stochastic neighbour embedding
(t-SNE) overview of identity and activity of chimaeric antigen receptor (CAR) and non-CAR immune cells over time after CAR-T-cell administra-
tion, analyzed with the OMIQ software in order to obtain the t-SNE maps. Shown is one representative patient. (a) Overview of non-CAR immune
cells and subsets and their activated and non-activated states. Activated CD4+ and CD8+ cells, cytolytic natural killer (NK) cells and classical mono-
cytes are revealed through combinatorial expression of respectively HLA-DR and CD69, CD16/CD56/NKp30 and CD14/CD16 over time after admin-
istration; the mean percentage of activated cells in the different populations is indicated. Activation reached a maximum on Day 7 or 14 and then
decreased with time. Cells are manually coloured according to their immune cell lineage. (b, c, d) Overview of CAR-T-cell activation and antigen-ex-
perience state (b) Activation profiles of CAR-T cells: the expression of HLA-DR and CD69 shows that almost all CAR- T cells were activated during
the 14 days following administration (percentages over time indicate CAR-T cells expressing HLA-DR and CD69). Black identifies unsigned (non-ac-
tivated) CAR-T cells isolated by manual gating (c) As a result of their activation, most of the CAR T cells were senescent, as shown by their CD57
expression; percentage of senescent cells over time is indicated (d) The differential expression patterns of CCR7 and CD45 RA identify the maturation
and antigen-experience states of CAR-T cells, such as na€ıve, effector, effector memory and central memory (left panel). Most CAR-T cells are effector
memory T cells; the percentage of both TEM and TEMRA cells is indicated. (B) Clustering of CAR and non-CAR immune cells. (Left) Clustering.
Data from blood samples collected on day 14 after CAR-T-cell administration were pooled, clustered and automatically annotated with the OMIQ
software offering an overview of highly expressed markers on specific populations (n = 5). (Right) PD1 expression on CAR-T cells over time. Note
the high expression of PD1 until day 28 on CD8+ CAR-T cells. [Colour figure can be viewed at wileyonlinelibrary.com]

the identification of the main immune cell populations and sub- Maxpar panel and flow cytometry strongly correlated for the
sets together with automatic data analyses remained valid using detection and quantification of immune cell subsets (r = 097,
this customized panel. As shown in Fig 1C, the customized P < 005; confirmed by the Bland–Altman method).12

ª 2021 British Society for Haematology and John Wiley & Sons Ltd
British Journal of Haematology, 2021, 194, 779–792

97
Correspondence

CAR-T-cell administration. This suggested that haematologic

Customized mass panel-supported
toxicity following CAR-T-cell infusions could be due to
comprehensive immunomonitoring after CAR-T
immune-mediated bone marrow failure as described in spo-
cell administration
radic aplastic anaemia and provide orientations for adapted
The combination of antibodies in our customized Maxpar therapeutic strategies.
panel allowed characterization of CAR and non-CAR Together, we found that mass cytometry-based platform
immune cells and their functional states, in a single pass and offers a powerful and reliable tool for comprehensive
over time (Fig 2A). Data were analyzed via the Maxpar Path- description of CAR-T and environmental immune cell sub-
setter software (Gemstone) which generates a visual display sets and opens new opportunities to identify immune signa-
of high-dimensional data. tures and support personalization of therapeutic strategy to
We present here preliminary observations on peripheral enhance CAR-T-cell efficacy.
blood of five patients followed over time after CAR-T-cell
administration. Kinetics of CAR and non-CAR immune cell
activation is poorly known and Fig 2A, panels a and b give a
Acknowledgements
simple and comprehensive picture of their massive activa- We thank B�en�edicte Hoareau-Coudert et Ang�elique Vinit for
tion. Almost all (92 4%) CAR-T cells, mainly CD8+, were technical support and Cl�ement Dumas for help in mass cyto-
activated on day 7 as demonstrated by HLA-DR and CD69 metric analysis.
expression (Fig 2A, panel b). Activation decreased from
day 14 but activated CAR-T could be detected up to five
months post administration. Exhaustion and senescence are
Funding
critical dysfunctional states impacted by persistent stimula- This work was supported by grants from the Soci�et�e
tion and a high level of senescence was observed in CAR-T Francaise de Greffe de Moelle and Association Capucine.
cells (mean, 65 5% from day 14 to 28; Fig 2A, panel c).
Among non-CAR immune cells (Fig 2A, panel a),
macrophages/monocytes have been reported as key mediators
Author contributions
in CAR-T-related toxicities; here almost all monocytes AC and CP performed mass cytometric analyses and wrote
demonstrated an inflammatory profile of M1 monocytes the paper. MC developed the mass-tagged CAR-T-cell anti-
CD14++C16 (mean, 93 4%) on days 7 and 14. The com- body. ET analyzed qPCR data. CB discussed results. DR-W
binatorial expression of HLA-DR and CD69, and CD16/ and SNG took patients in care and discussed results. EL and
CD56/NKp30, demonstrated the activated states of TCD8+ AG supervised the cytometric analyses and discussed results.
lymphocytes (mean, 82 7%) and cytolytic NK cells (mean, FN conceived of and designed the experiments, supervised
65 4%) respectively on days 7 and 14. the research and wrote the paper.
It is now clear that less differentiated lymphocytes are
more effective in the transfer of immunity for adoptive cell
therapy and our approach also provided information regard-
Conflict of interest
ing maturation and antigen-experience states (na€ıve, effector, The authors declare to have no potential conflicts of interest
central memory) of CAR and non-CAR cells by evaluating regarding the present work.
the differential expression patterns of CCR7 and CD45RA.13
Aur�elien Corneau1,*
In our small group of patients, the effector memory state of
Christophe Parizot2,3,*
CAR-T cells varied from 64% to 95% (Fig 2A, panel d).
Mustapha Cherai2,3
Finally, this platform allowed the clustering of CAR and
Eve Todesco4
non-CAR immune cells and offered an overview of highly
Catherine Blanc1
expressed markers on specific subpopulations (Fig 2B). We
Elena Litvinova2
could note a high expression of PD1 until day 28 on CD8+
St�ephanie Nguyen5
CAR-T cells (mean, 53 4%) suggesting that CAR-T cells
Damien Roos-Weil5
could in part circumvent the PD-L1/PD1 brake.14 The
Am�elie Guihot2,3
expression of PD1 and some chemokine receptors on CAR-T
Francoise Norol5
cells could be a sensitive biomarker as suggested by the three 1
Sorbonne Universit�e (Univ. Paris 06), UMS37-PASS, Plateforme de
patients who achieved complete or partial responses and this
cytom�etrie, Hôpital Piti�e-Salpêtri�ere, 2Assistance Publique-Hôpitaux de
could support further investigations.
Paris (AP-HP), Hôpital Piti�e-Salpêtri�ere, D�epartement d’Immunologie
This integrated method could also track signatures for
F-75013, 3Sorbonne Universit�e (Univ. Paris 06), INSERM U1135, Cen-
personalized therapeutic strategies. In Figure S2, we also
tre d’Immunologie et des Ma-ladies Infectieuses (CIMI-Paris), Hôpital
report the real-time characterization of an activated popula-
Piti�e-Salpêtri�ere, 4Assistance Publique-Hôpitaux de Paris (AP-HP),
tion of Th17 lymphocytes in blood and bone marrow sam-
Hôpital Piti�e-Salpêtri�ere, Laboratoire de Virologie, F-75013 and
ples of a patient that experienced a profound aplasia after

ª 2021 British Society for Haematology and John Wiley & Sons Ltd
British Journal of Haematology, 2021, 194, 779–792

98
Correspondence

5
Assistance Publique-Hôpitaux de Paris (AP-HP), Service 2. Mueller KT, Maude SL, Porter DL, Frey N, Wood P, Han X, et al. Cellular
d’H�ematologie, Hôpital Piti�e Salpêtri�ere, Paris, F-75013, France. kinetics of CTL019 in relapsed/refractory B-cell acute lymphoblastic leuke-
mia and chronic lymphocytic leukemia. Blood. 2017;130:2317–25.
E-mail: [email protected]
3. Shah NN, Fry T. Mechanisms of resistance to CAR T cell therapy. Nat Rev
*AC and CP contributed equally to this work. Clin Oncol. 2019;16:372–85.
4. Fraietta JA, Lacey SF, Orlando EJ, Pruteanu-Malinici I, Gohil M, Lundh S,
Keywords: chimaeric antigen receptor-T cells, immunomonitoring, et al. Determinants of response and resistance to CD19 chimeric antigen
mass cytometry receptor (CAR) T cell therapy of chronic lymphocytic leukemia. Nat Med.
2018;24:563–71.
5. Du M, Hari P, Hu Y, Mei H. Biomarkers in individualized management
First published online 26 May 2021
of chimeric antigen receptor T cell therapy. Biomarker. 2020;8:1–13.
doi: 10.1111/bjh.17551 6. Gadalla R, Noamani B, MacLeod BL, Dickson RJ, Guo M, Xu W, et al.
Validation of CyTOF against flow cytometry for immunological studies
and monitoring of human cancer clinical trials. Front Oncol. 2019;9:415.
Supporting Information 7. Hartmann FJ, Babdor J, Gherardini PF, Amir E-A, Jones K, Sahaf B, et al.
Comprehensive immune monitoring of clinical trials to advance human
Additional supporting information may be found online in immunotherapy. Cell Rep. 2019;28:819–31.
the Supporting Information section at the end of the article. 8. Bagwell CB, Hunsberger B, Hill B, Herbert D, Bray C, Selvanantham T,
Table SI. Characteristics of the patients. et al. Multi-site reproducibility of a human immunophenotyping assay in
Fig S1. The CAR-T-cell detection reagent from Miltenyi whole blood and peripheral blood mononuclear cells preparations using
CyTOF technology coupled with Maxpar Pathsetter, an automated data
Biotec (Bergisch Gladbach, Germany) provides better dis-
analysis system. Cytometry B Clin Cytom. 2020;98:146–60.
crimination in comparison with the Acrobiosystems (New- 9. Lee DW, Santomasso BD, Locke FL, Ghobadi A, Turtle CJ, Brudno JN,
ark, DE, USA) reagent for CD19+ CAR-T-cell determination. et al. ASBMT consensus grading for cytokine release syndrome and neuro-
Fig S2. Signature associated with post CAR-T cells aplasia logic toxicity associated with immune effector cells. Biol Blood Marrow
in a given patient. Real-time characterization of activated Transplant. 2019;25:625–38.
10. De Oliveira SN, Wang J, Ryan C, Morrison SL, Kohn DB, Hollis RP. A
populations of Th17 and CD69 lymphocytes and regulatory
CD19/Fc fusion protein for detection of anti-CD19 chimeric antigen
T lymphocytes (TReg) in the blood and bone marrow of a receptors. J Transl Med. 2013;11:23.
patient displaying profound aplasia on day 60 after CAR-T- 11. Hauser JR, Hong H, Babady NE, Papanicolaou GA, Tang Y-W. False-posi-
cell administration (left and middle panels). Note that in tive results for human immunodeficiency virus type 1 nucleic acid amplifi-
blood, the Th17 cell population is present at a high level cation testing in chimeric antigen receptor T cell therapy. J Clin Microbiol.
2019;58:e01420–e1519.
compared to that in a panel of patients who rapidly recov-
12. Appay V, Dunbar PR, Callan M, Klenerman P, Gillespie GMA, Papagno L,
ered following CAR-T-cell infusion (right panel). Black indi- et al. Memory CD8+ T cells vary in differentiation phenotype in different
cates unsigned cell populations. persistent virus infections. Nat Med. 2002;8:379–85.
13. McLellan AD, Ali Hosseini Rad SM. Chimeric antigen receptor T cell per-
sistence and memory cell formation. Immunol Cell Biol. 2019;97:664–74.
References 14. Dong E, Yue X-Z, Shui L, Liu B-R, Li Q-Q, Yang Y, et al. IFN-c surmounts
PD-L1/PD1 inhibition to CAR-T cell therapy by upregulating ICAM-1 on
1. June CH, Sadelain M. Chimeric antigen receptor therapy. N Engl J Med.
tumor cells signal transduction and targeted. Therapy. 2021;6:20.
2018;379:64–73.

ª 2021 British Society for Haematology and John Wiley & Sons Ltd
British Journal of Haematology, 2021, 194, 779–792

99
TM

Krok 1 - 2020 (Oct 30) (General Medicine) - Eneutron
No ratings yet
Krok 1 - 2020 (Oct 30) (General Medicine) - Eneutron
44 pages
AACR 2022 Proceedings: Part A Online-Only and April 10
From Everand
AACR 2022 Proceedings: Part A Online-Only and April 10
CTI Meeting Technology
No ratings yet
AACR 2018 Proceedings: Abstracts 3028-5930
From Everand
AACR 2018 Proceedings: Abstracts 3028-5930
CTI Meeting Technology
No ratings yet
Fast Facts: Molecular Profiling in Solid Tumors
From Everand
Fast Facts: Molecular Profiling in Solid Tumors
Parneet Cheema
No ratings yet
Immunophenotyping Techniques for Biomedical Scientists: Continuing Professional Development in Pathology For Medical Laboratory Professionals
From Everand
Immunophenotyping Techniques for Biomedical Scientists: Continuing Professional Development in Pathology For Medical Laboratory Professionals
Dr Lydia Taiwo
No ratings yet
Cytometry Pt A - 2022 - Jaimes - Full spectrum flow cytometry and mass cytometry A 32%E2%80%90marker panel comparison
No ratings yet
Cytometry Pt A - 2022 - Jaimes - Full spectrum flow cytometry and mass cytometry A 32%E2%80%90marker panel comparison
18 pages
AACR 2018 Proceedings: Abstracts 1-3027
From Everand
AACR 2018 Proceedings: Abstracts 1-3027
CTI Meeting Technology
No ratings yet
AACR 2019 Proceedings: Abstracts 1-2748
From Everand
AACR 2019 Proceedings: Abstracts 1-2748
CTI Meeting Technology
No ratings yet
AACR 2016: Abstracts 2697-5293
From Everand
AACR 2016: Abstracts 2697-5293
American Association for Cancer Research (AACR)
No ratings yet
Bold Medical Leaps
From Everand
Bold Medical Leaps
Nora Franklin
No ratings yet
A Compilation of Cellular and Tissue Xenotransplantation Studies (1838-2022): A Promising Approach for the Treatment of Diseases
From Everand
A Compilation of Cellular and Tissue Xenotransplantation Studies (1838-2022): A Promising Approach for the Treatment of Diseases
Prof Dr Mike KS Chan
No ratings yet
Fast Facts: Managing Immune-Related Adverse Events in Oncology
From Everand
Fast Facts: Managing Immune-Related Adverse Events in Oncology
Helen Westman
No ratings yet
AACR 2017 Proceedings: Abstracts 1-3062
From Everand
AACR 2017 Proceedings: Abstracts 1-3062
American Association for Cancer Research (AACR)
No ratings yet
AACR 2017 Proceedings: Abstracts 3063-5947
From Everand
AACR 2017 Proceedings: Abstracts 3063-5947
American Association for Cancer Research (AACR)
No ratings yet
Clinical Flow Cytometry: A Transition in Utilization
No ratings yet
Clinical Flow Cytometry: A Transition in Utilization
19 pages
Crossing the Valley of Death with Advanced Cancer Therapy
From Everand
Crossing the Valley of Death with Advanced Cancer Therapy
Akseli Hemminki
No ratings yet
Role of Nanotechnology in Cancer Therapy
From Everand
Role of Nanotechnology in Cancer Therapy
Priya Patel
No ratings yet
Transfer Factors: Properties, Mechanism of Action and Its Clinical Applications
From Everand
Transfer Factors: Properties, Mechanism of Action and Its Clinical Applications
Prof Dr Mike KS Chan
No ratings yet
Immunotherapy Cancer Battle: Innovative Approaches for Success
From Everand
Immunotherapy Cancer Battle: Innovative Approaches for Success
Azhar ul Haque Sario
No ratings yet
Personalized Immunotherapy for Tumor Diseases and Beyond
From Everand
Personalized Immunotherapy for Tumor Diseases and Beyond
PublishDrive
No ratings yet
AACR 2019 Proceedings: Abstracts 2749-5314
From Everand
AACR 2019 Proceedings: Abstracts 2749-5314
CTI Meeting Technology
No ratings yet
s41598-023-32608-2
No ratings yet
s41598-023-32608-2
11 pages
Gene Editing Advances
From Everand
Gene Editing Advances
Zara Sagan
No ratings yet
Jove Protocol 66729 Flow Cytometry Based Monitoring of Cancer Immunity Cycle
No ratings yet
Jove Protocol 66729 Flow Cytometry Based Monitoring of Cancer Immunity Cycle
20 pages
Immunotherapy Cancer Battle: Innovative Approaches for Success, Second Edition
From Everand
Immunotherapy Cancer Battle: Innovative Approaches for Success, Second Edition
Azhar ul Haque Sario
No ratings yet
AACR 2022 Proceedings: Part B April 11-13
From Everand
AACR 2022 Proceedings: Part B April 11-13
CTI Meeting Technology
No ratings yet
Stem Cells Between Regeneration and Tumorigenesis
From Everand
Stem Cells Between Regeneration and Tumorigenesis
Cristiana Tanase
No ratings yet
Fast Facts: Peripheral T-cell Lymphoma: Unraveling the Complexities of Diagnosis and Management
From Everand
Fast Facts: Peripheral T-cell Lymphoma: Unraveling the Complexities of Diagnosis and Management
Francine Foss
No ratings yet
Diagnostic Problems in Tumors of Head and Neck: Selected Topics: DIAGNOSTIC PROBLEMS IN TUMOR PATHOLOGY SERIES, #1
From Everand
Diagnostic Problems in Tumors of Head and Neck: Selected Topics: DIAGNOSTIC PROBLEMS IN TUMOR PATHOLOGY SERIES, #1
Arun Chitale
5/5 (1)
Multiple Sclerosis: Practical Comprehensive Essentials
From Everand
Multiple Sclerosis: Practical Comprehensive Essentials
Mike K.S. Chan
3/5 (1)
Applications in Cytometry Now CyTOF and Neuroscience (FLDM-01351 Rev 01) 202405
No ratings yet
Applications in Cytometry Now CyTOF and Neuroscience (FLDM-01351 Rev 01) 202405
12 pages
Biotechnology Breakthroughs - The Innovations Changing Medicine
From Everand
Biotechnology Breakthroughs - The Innovations Changing Medicine
Julia Evans
No ratings yet
Thomas' Hematopoietic Cell Transplantation
From Everand
Thomas' Hematopoietic Cell Transplantation
Frederick R. Appelbaum
No ratings yet
Fast Facts: Managing immune-related Adverse Events in Oncology: Early recognition, prompt intervention, effective management
From Everand
Fast Facts: Managing immune-related Adverse Events in Oncology: Early recognition, prompt intervention, effective management
B.L. Rapoport
No ratings yet
Understanding Hodgkin Lymphoma. A Guide For Patients, Survivors, and Loved Ones. October 2017
From Everand
Understanding Hodgkin Lymphoma. A Guide For Patients, Survivors, and Loved Ones. October 2017
Lymphoma Research Foundation
No ratings yet
Fast Facts: Waldenström Macroglobulinemia
From Everand
Fast Facts: Waldenström Macroglobulinemia
Christian Buske
No ratings yet
Handbook of Anti-Aging Medicine
From Everand
Handbook of Anti-Aging Medicine
Prof Dr Mike KS Chan
No ratings yet
Fast Facts: Blastic Plasmacytoid Dendritic Cell Neoplasm: Shedding Light on a Rare Disease
From Everand
Fast Facts: Blastic Plasmacytoid Dendritic Cell Neoplasm: Shedding Light on a Rare Disease
Eric Deconinck
No ratings yet
Article Virginia Litwin Using Spectral Flow Cytometry For Car T Cell Clinical Trials
No ratings yet
Article Virginia Litwin Using Spectral Flow Cytometry For Car T Cell Clinical Trials
12 pages
Current Trends in the Identification and Development of Antimicrobial Agents
From Everand
Current Trends in the Identification and Development of Antimicrobial Agents
PublishDrive
No ratings yet
Diagnostic Problems in Tumors of Central Nervous System: Selected Topics
From Everand
Diagnostic Problems in Tumors of Central Nervous System: Selected Topics
Arun Chitale
No ratings yet
Stem Cells in Regenerative Medicine: Carpe Diem – Carpe Vitam!
From Everand
Stem Cells in Regenerative Medicine: Carpe Diem – Carpe Vitam!
Mike K.S. Chan
2.5/5 (3)
AACR 2016: Abstracts 1-2696
From Everand
AACR 2016: Abstracts 1-2696
American Association for Cancer Research (AACR)
No ratings yet
Fast Facts: Neuroblastoma
From Everand
Fast Facts: Neuroblastoma
Jennifer Foster
No ratings yet
Diagnostic Problems in Tumors of Endocrine System: Selected Topics: DIAGNOSTIC PROBLEMS IN TUMOR PATHOLOGY SERIES, #5
From Everand
Diagnostic Problems in Tumors of Endocrine System: Selected Topics: DIAGNOSTIC PROBLEMS IN TUMOR PATHOLOGY SERIES, #5
Arun Chitale
No ratings yet
Fast Facts: Cutaneous T-cell Lymphoma
From Everand
Fast Facts: Cutaneous T-cell Lymphoma
Paul L. Haun
No ratings yet
Biomarkers in Medicine
From Everand
Biomarkers in Medicine
PublishDrive
No ratings yet
Genetics Research And Textbook 3
From Everand
Genetics Research And Textbook 3
Aliasghar Tabatabaei Mohammadi
No ratings yet
Immunotherapy in Transplantation: Principles and Practice
From Everand
Immunotherapy in Transplantation: Principles and Practice
Bruce Kaplan
No ratings yet
2 Part 2
No ratings yet
2 Part 2
22 pages
Exploring Surgical Pathology: A Practical Guide to Specimen Dissection and Analysis
From Everand
Exploring Surgical Pathology: A Practical Guide to Specimen Dissection and Analysis
Pasquale De Marco
No ratings yet
68th AACC Annual Scientific Meeting Abstract eBook
From Everand
68th AACC Annual Scientific Meeting Abstract eBook
American Association for Clinical Chemistry (AACC)
No ratings yet
Thomas' Hematopoietic Cell Transplantation: Stem Cell Transplantation
From Everand
Thomas' Hematopoietic Cell Transplantation: Stem Cell Transplantation
Stephen J. Forman
No ratings yet
Comprehensive Characterization of The Multiple Mye
No ratings yet
Comprehensive Characterization of The Multiple Mye
32 pages
Standardization of Suspension and Imaging Mass Cyt
No ratings yet
Standardization of Suspension and Imaging Mass Cyt
36 pages
Nutritional Genetics Part 5: Diseases C
From Everand
Nutritional Genetics Part 5: Diseases C
Ronald Steriti
No ratings yet
Molecular Targets and Cancer Therapeutics (Part 2)
From Everand
Molecular Targets and Cancer Therapeutics (Part 2)
PublishDrive
No ratings yet
Medical Science Leaps
From Everand
Medical Science Leaps
Eleanor Hawking
No ratings yet
Frontiers in Clinical Drug Research - Anti-Cancer Agents: Volume 4
From Everand
Frontiers in Clinical Drug Research - Anti-Cancer Agents: Volume 4
Atta-ur-Rahman
No ratings yet
Fast Facts: The Essentials of Cytopathology
From Everand
Fast Facts: The Essentials of Cytopathology
Peter Dalquen
No ratings yet
Blog Post
No ratings yet
Blog Post
6 pages
jtd-06-06-E133
No ratings yet
jtd-06-06-E133
4 pages
bmjebm-2022-112126
No ratings yet
bmjebm-2022-112126
6 pages
FeedingTubeMeds_DSF
No ratings yet
FeedingTubeMeds_DSF
18 pages
D-Fever in Children With Chemotherapy-Induced Neutropenia - UpToDate
No ratings yet
D-Fever in Children With Chemotherapy-Induced Neutropenia - UpToDate
41 pages
Erwinase10,000 IU
No ratings yet
Erwinase10,000 IU
5 pages
2-CYP17 Inhibitors For Prostate Cancer Therapy
No ratings yet
2-CYP17 Inhibitors For Prostate Cancer Therapy
18 pages
1-Guidelines For Safe Medication Adminstration (Non Parentral Medication Adminstration)
No ratings yet
1-Guidelines For Safe Medication Adminstration (Non Parentral Medication Adminstration)
6 pages
Bilirubin Met
No ratings yet
Bilirubin Met
27 pages
Cancer Screening Guidelines Lack Information On Harms - NCI
No ratings yet
Cancer Screening Guidelines Lack Information On Harms - NCI
10 pages
Osapnea
No ratings yet
Osapnea
10 pages
77-Liver and Bile Duct Cancer - CHAPTER 77
No ratings yet
77-Liver and Bile Duct Cancer - CHAPTER 77
42 pages
DASH Eating Plan - NHLBI, NIH
No ratings yet
DASH Eating Plan - NHLBI, NIH
7 pages
Aasld Practice Guidance On Prevention, Diagnosis,.441
No ratings yet
Aasld Practice Guidance On Prevention, Diagnosis,.441
104 pages
1'-Abiraterone Acetate - Spironolactone
No ratings yet
1'-Abiraterone Acetate - Spironolactone
1 page
Thrombexx Ampoule Leaflet
No ratings yet
Thrombexx Ampoule Leaflet
1 page
IB Biology B2.2
No ratings yet
IB Biology B2.2
3 pages
Cap 88-7 Fanaroff & Martin's Neonatal Perinatal Medicine (11-22)
No ratings yet
Cap 88-7 Fanaroff & Martin's Neonatal Perinatal Medicine (11-22)
12 pages
CLS Aipmt 16 17 XIII Bot Study Package 2 SET 1 Chapter 6
100% (2)
CLS Aipmt 16 17 XIII Bot Study Package 2 SET 1 Chapter 6
37 pages
Developmental Biology Test Questions
No ratings yet
Developmental Biology Test Questions
73 pages
Pathophysiology of Stroke
100% (3)
Pathophysiology of Stroke
22 pages
Class 10 Biology Test Paper 2
No ratings yet
Class 10 Biology Test Paper 2
2 pages
CHAPTER 3 - The Cellular Level of Organization
No ratings yet
CHAPTER 3 - The Cellular Level of Organization
82 pages
10th Biology Important Diagrams
No ratings yet
10th Biology Important Diagrams
8 pages
Common Laboratory and Diagnotics Examination
No ratings yet
Common Laboratory and Diagnotics Examination
10 pages
Flyer 2020 V2 Pediatrics
No ratings yet
Flyer 2020 V2 Pediatrics
2 pages
First Year M.B.B.S. Examination March / April - 2019 Anatomy Paper 1
No ratings yet
First Year M.B.B.S. Examination March / April - 2019 Anatomy Paper 1
3 pages
Week 8 Learning Activity
No ratings yet
Week 8 Learning Activity
4 pages
Kode Icd 10
No ratings yet
Kode Icd 10
2 pages
Learning Module in General Biology 1: First Semester, S.Y. 2020 - 2021
No ratings yet
Learning Module in General Biology 1: First Semester, S.Y. 2020 - 2021
6 pages
Mouth Anatomy
No ratings yet
Mouth Anatomy
12 pages
Questions Renal Physiology Find The Mode of Action of A Diuretic Drug
0% (1)
Questions Renal Physiology Find The Mode of Action of A Diuretic Drug
3 pages
The Fetal Skull: Bones of The Vault
100% (1)
The Fetal Skull: Bones of The Vault
3 pages
physiology test bank
No ratings yet
physiology test bank
24 pages
Basolo PDF Sonn
No ratings yet
Basolo PDF Sonn
11 pages
Finals in Forensic 1
No ratings yet
Finals in Forensic 1
9 pages
Respiratory System
No ratings yet
Respiratory System
6 pages
Immunity 1 Notes
No ratings yet
Immunity 1 Notes
43 pages
Approach To Head CT
No ratings yet
Approach To Head CT
72 pages
PERFRMANCE TASK 3 and 4 Blood Grouping
No ratings yet
PERFRMANCE TASK 3 and 4 Blood Grouping
5 pages
The Language of Anatomy P1
No ratings yet
The Language of Anatomy P1
3 pages
Sieve Elements - Comparative Structure, Induction and Development (1990)
No ratings yet
Sieve Elements - Comparative Structure, Induction and Development (1990)
310 pages
Wound Healing: Healing by First Intention (Primary Union)
No ratings yet
Wound Healing: Healing by First Intention (Primary Union)
31 pages
Toaz - Info Imaginefx Presents How To Draw and Paint Anatomy Uk PR 55f8581af53dc57 2
No ratings yet
Toaz - Info Imaginefx Presents How To Draw and Paint Anatomy Uk PR 55f8581af53dc57 2
3 pages
Difference Between Diploblastic and Triploblastic: Why Men Leave Women They Love
No ratings yet
Difference Between Diploblastic and Triploblastic: Why Men Leave Women They Love
9 pages

Immune Landscape v11

Uploaded by

Immune Landscape v11

Uploaded by

Surveying the

Immune Landscape of Cancer

37 Mass cytometry: a robust platform for

A phase1b/2 study of azacitidine with

Assessment of the immune landscapes of advanced ovarian

Simone Pisano1,2 Stefania Lenna1 Gareth D. Healey2 Fereshteh Izardi2

1Department of Nanomedicine, Houston

Clin. Transl. Med. 2021;11:e551. wileyonlinelibrary.com/journal/ctm2

as a diagnostic tool for this malignancy, with the opportunity to concomitantly

1 INTRODUCTION total absence) of CD8+ T cells within the TME,12 or the

3 groups (n = 5 mice per group) and injected intraperi-

toneally with 5 × 106 , 1 × 107 , or 1.5 × 107 ID8-Luc/GFP cells

in 200 μL of PBS. Cells were injected into the lower right

1.7 × 104 and 9.7 × 104 photons/s, respectively.

Seventy days after tumor cell injection with 1 × 107

onto 50% Percoll and centrifuged at 620 × g without the

injection of 1 × 107 or 1.5 × 107 cells. Statistical difference

pared to the day of injection (P < 0.001). This difference

became more marked after day 23 days in 5 × 106 group and

after 37 days in 1 × 107 group. No statistically significant

intragroup differences were recorded in 1.5 × 107 group.

analyses, respectively. Data with a P < 0.05 were consid-

expressed as the mean ± standard deviation (SD).

3 RESULTS 3.3 Immunohistochemistry of immune

After ID8-Luc/GFP cells (5 × 106 , 1 × 107 or 1.5 × 107 ) were

1 × 107 cells showed a 13.6 ± 9 fold increase in tumor

1 (P < 0.001), as quantified by IVIS (Figure 1A). Mice

injected with 5 × 106 or 1.5 × 107 cells showed a 8.2 ± 4

fold and 5.6 ± 43.9 fold increase in tumor growth, respec-

and CTL cells (4.6% ± 3.1% and 4.85% ± 3.4%, respectively)

± 9.6%), CD8+ T cells (38.5% ± 4.5%), and CD4+ T cells

(20.7% ± 3.5%), with myeloid immune cells, including

T cells (Ly6C+/CD44+, 23.1% ± 12.1%), (ii) T helper cells

(IFNγ+/CD4+, 22.2% ± 15.3%), and (iii) cytotoxic T lym-

phocytes (CTL, IFNγ+/CD8+, 12.5% ± 9.05%). In addition,

low expression of PD1 in 8.03% ± 7.4% of CD4+ T cells and

2.3% ± 1.5% of CD8+ T cells was noted indicating a low

were 61.9% ± 24.3% and 69.7% ± 28.5%, respectively. The

injected (between 1 × 106 and 1 × 107 ),32,36–39 and different

Luc/GFP cells (5 × 106 , 1 × 107 , or 1.5 × 107 cells) for

CD8+ CD4+ T NK Dendritic

of >30 selected antigen markers. Moreover, CyTOF “nar-

In this work, we provide, for the first time, a compre- AVA I L A B I L I T Y O F D A T A A N D

Compr Cancer Netw. 2019;17(8):896-909. https://doi.org/10.6004/ Immunother. 2001;24:392-407. https://doi.org/10.1097/

infiltrating programmed death receptor-1 + dendritic cells

65. Xu Y, Chen L, Xu B, et al. Higher numbers of T-bet + tumor- 2007;109(9):3812-3819. https://doi.org/10.1182/blood-2006-07-

The detection of tumor-specific T cells in solid tumors is integral to interrogate endoge-

Keywords: CD39  CD103  CD137  PD-1  tumor-infiltrating lymphocytes

Correspondence: Daniel J. Powell Jr.

© 2021 Wiley-VCH GmbH www.eji-journal.eu

© 2021 Wiley-VCH GmbH www.eji-journal.eu

© 2021 Wiley-VCH GmbH www.eji-journal.eu

© 2021 Wiley-VCH GmbH www.eji-journal.eu

© 2021 Wiley-VCH GmbH www.eji-journal.eu

© 2021 Wiley-VCH GmbH www.eji-journal.eu

in CD4+ TILs. CD8+ TILs contained greater frequencies of cells

Tumor-specific TILs display a phenotype indicative of

Since our findings indicate that CD137+ TILs express effec-

© 2021 Wiley-VCH GmbH www.eji-journal.eu

© 2021 Wiley-VCH GmbH www.eji-journal.eu

© 2021 Wiley-VCH GmbH www.eji-journal.eu

© 2021 Wiley-VCH GmbH www.eji-journal.eu

© 2021 Wiley-VCH GmbH www.eji-journal.eu

© 2021 Wiley-VCH GmbH www.eji-journal.eu

© 2021 Wiley-VCH GmbH www.eji-journal.eu

Identification of the immune checkpoint signature of

Keywords: immune checkpoint, immunotherapy, mass cytometry,

patients, largely because of the fact that

The immune checkpoint landscape of MM The immune checkpoint atlas of MM BM

the BM were also increased in MM patients indicated immune checkpoint protein-expressing

Affiliated Hospital of Sun Yat-sen University. All protocols

REFERENCES malignancy: preliminary results of a phase Ib study. J

A Phase 1b/2 Study of Azacitidine With PD-L1 Antibody

Cancer October 15, 2021

Cancer October 15, 2021

Cancer October 15, 2021

Keywords: CD39 CD103 CD137 PD-1 tumor-infiltrating lymphocytes