Available via license: CC BY-NC-ND 4.0
Content may be subject to copyright.
Technology Explained
A user's perspective on GeoMx
TM
digital spatial profiling
Trieu My Van
*
,
1
, Christian U. Blank
Netherlands Cancer Institute, Amsterdam, The Netherlands
ARTICLE INFO
Keywords:
NanoString
DSP
Region of interest
Multiplex
IHC
Translational research
ABSTRACT
Characterization of spatial protein expression for multiple targets from a single tissue is difficult to perform,
especially due to the limitations of multiplex immunohistochemistry and tissue heterogeneity. Therefore, a new
technology is required that permits detailed and simultaneous expression profiling of proteins within a defined
region of interest (ROI). To address this unmet need, NanoString Technologies developed a new technology,
GeoMx
TM
digital spatial profiling (DSP), which currently enables simultaneous and guided detection of up to 40
antibodies (probes) from a single formalin-fixed paraffin-embedded (FFPE) tissue. DSP probes are tagged with
unique photocleavable DNA oligos that are released after guided ultraviolet exposure in specific ROIs. Digital
quantification of the released oligos by NanoString's nCounter®system provides a detailed expression profile of
proteins within these discrete ROIs. In this article, we will describe our experience with the GeoMx DSP platform
using cancer FFPE tissues. These expression profiles will provide better characterization and understanding of
tumor heterogeneity and the tumor micro-environment, enabling the improvement of patient therapy and the
identification of potential biomarker signatures. The purpose of this article is to offer potential future users an
independent insight into the DSP platform and a comprehensive idea of usability, including advantages and
current limitations of the technology based on our current experience with the beta version of NanoString's DSP
platform as part of the DSP beta-testing program. The GeoMx
TM
Digital Spatial Profiling (DSP) platform is a non-
destructive technique for regional in-depth protein expression profiling. Using oligonucleotide detection tech-
nologies, the GeoMx
TM
DSP enables simultaneous high-level multiplexing on a single FFPE tissue. Here, we focus
on our current experience derived from our biomarker research using the beta version of the DSP instrument.
Background
Regional and quantitative protein signature analysis of clinical
formalin-fixed paraffin-embedded (FFPE) tissues has proved to be diffi-
cult to perform in many research areas. In particular, in the field of
immuno-oncology (IO), detailed characterization of immune cell subsets
within a tumor area would improve our understanding of antitumor
immunity and resistance to immune checkpoint inhibition. Tumors are
highly heterogeneous in their architecture, (immune) cell composition,
abundance and distribution. Understanding tumor and immune cell co-
localization may also be important for biomarker identification and
precision immunotherapy [1–5]. At present, the CD8þT cell:regulatory T
cell ratio is an improved biomarker compared with CD8 expression alone
[6–11]. Furthermore, characterizing co-expression of co-stimulatory and
inhibitory receptors would improve therapeutic strategies. For example,
characterization of inhibitory markers expressed on T cells (e.g. LAG-3
and TIM-3) in the tumor micro-environment may impact the decision
for new (combination) therapies [12,13]. Detailed tissue analysis is
highly necessary but involves a laborious staining procedure and is
hampered by limited patient tissue samples. Therefore, multiplex
immunohistochemistry (IHC) will provide a more comprehensive insight
into the interaction and crosstalk between tumor and immune cells
within the tumor micro-environment.
Multiplexing more than five antibodies has largely been restricted by
the spectral overlap of available fluorophores or chromogens. Moreover,
current multiplex IHC tools provide objective information about the
presence and histological location of immune cells, but neglect detailed
phenotypical description. Due to this limitation, additional flow cytom-
etry analysis is required. To advance standard IHC-based multiplex tissue
analyses, NanoString Technologies (Seattle, WA) developed a novel im-
aging and tissue-sampling platform: GeoMx
TM
digital spatial profiling
(DSP). This is a high-level multiplexing technique that provides infor-
mation about the presence and histological location of immune cell
subsets, and detailed characterization of their activation, differentiation
* Corresponding author. Trieu M. Van, Netherlands Cancer Institute, Amsterdam, The Netherlands. Tel.: þ31 20 512 2066.
E-mail address: t.van@nki.nl (T.M. Van).
1
Postal Adress: Plesmanlaan 121, 1066 CX Amsterdam
Contents lists available at ScienceDirect
Immuno-Oncology Technology
journal homepage: www.esmoiotech.org
https://doi.org/10.1016/j.iotech.2019.05.001
Available online 30 May 2019
2590-0188/©2019 Published by Elsevier Ltd on behalf of European Society for Medical Oncology. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Immuno-Oncology Technology 1 (2019) 11–18
and immune checkpoint expression [14].
The capability of DSP technology in the field of IO holds promise in
advancing the current standard analysis of tumor biopsies. Thus, this
article aims to give an insight into the DSP technology, the current rec-
ommended experimental set-up and an independent view on the ad-
vantages and limitations of the system based on our own experience and
research. It should be noted that all observations are derived from our
experience with the beta version of the DSP instrument, which we are
currently testing, and may not apply fully to the commercially available
instrument.
The working procedure of protein DSP
DSP technology simultaneously characterizes regional and quantita-
tive protein expression of up to 40 markers related to immune cell acti-
vation and tumor cell classification on a single FFPE tissue section [14].
The DSP procedure implements five non-destructive steps: a standard
FFPE tissue preparation step, a tissue incubation step with a mixture of
visualization markers (VMs) and DSP probes, an imaging and region of
interest (ROI) selection step, an ultraviolet (UV) exposure and oligo
collection step, and a quantification step on NanoString's nCounter®
system (Figure 1)[14]. In total, this procedure takes 3 days from tissue
slide preparation until data analysis, with a throughput of four slides per
day. Below, we will provide a more detailed review of our experiences
with the DSP platform and its advantages and limitations.
Tissue preparation
Tissue preparation for DSP analysis is comparable to standard IHC,
and is therefore an established procedure that does not involve additional
steps or reagents. A major advantage of DSP is the non-destructive
staining procedure. Tissue sections can be preserved after staining/
acquisition for further haematoxylin and eosin or chromogenic IHC
staining. Moreover, processed slides can be stored long term for addi-
tional DSP analysis. Further analysis will require additional antigen
retrieval and re-incubation with a cocktail of VMs and DSP probes.
Tissue incubation with VMs and DSP probes
For DSP analysis, the tissue sample requires an incubation step with a
master mix containing the VMs and DSP probes (Table 1). The VMs are
tissue-compartment-specificfluorescent-labeled antibodies and a DNA
marker used to visualize the tissue morphology. ROI selection from VMs
has the advantage of allowing researchers to select areas based on the
immune cell type of interest and the surrounding environment. The DSP
platform enables multispectral imaging but also minimizes emission-
spectral overlap by including four distinct light-emitting diodes. This
limits the morphological characterization of the tissue to three different
markers. Currently, there are no options within the DSP platform to
extend the number of VMs, and thus thorough selection of markers or the
use of serial tissue sections is required. Based on our experience,
combining more than four VMs by applying an intermediate bleaching
step is not recommended due to remaining tissue stain and the conse-
quent possibility of false-positive signals.
The tissue is also incubated with the DSP probe panel, which consists
of a fixed core panel that can be extended with two additional modules
(Table 1B). These DSP probes are not multiplex-limited by spectral res-
olution, as they are not fluorescently labelled but are labelled with bar-
code indexing oligos. These oligos are linked to each antibody via a UV-
cleavable linker modified from a previous concept [15]. During the beta
test phase, the DSP probe panel consists of 40 antibodies. Theoretically,
the DSP probe panel could be extended to a maximum of 96 antibodies
Visualization
marker
Figure 1. Protein digital spatial profiling (DSP) working procedure. Formalin-fixed paraffin-embedded tissue slide preparation involves incubation with an antibody
mixture which contains up to four visualization markers and 40 DSP probes. Following imaging, regions of interest (ROIs) are selected based on visualization of the
tissue. Sequential ultraviolet (UV) light exposure of each ROI results in the release of indexing oligos from the DSP probes, allowing their quantification on Nano-
String's nCounter®system.
Table 1
Overview of (A) visualization markers used currently to distinguish between the
tumor and immune cell compartment, and (B) digital spatial profiling protein
probes included in the core panel and the additional modules
A
Visualization markers Compartment
Syto13 DNA
S100B/Pmel17 or PanCK Tumor
CD45 Immune cell
CD3 Immune cell
B
Core panel Module 1 Module 2
Beta-2-microglobulin CD56 CD137 CD127
CD11c HLA-DR LAG3 CD25
CD20 SMA OX40L CD80
CD3 Fibronectin TIM-3 CD86
CD4 TGFB1 VISTA ICOS
CD45 PD-L1 ARG1 PD-L2
CD68 GZMB B7-H3 CD40
CD8 Ki-67 IDO1 CD40L
CTLA4 PD1 STING CD27
PanCK IgGs GITR CD44
T.M. Van, C.U. Blank Immuno-Oncology Technology 1 (2019) 11–18
12
due to the technical capability of the quantification process in the
nCounter system. For our projects, we extended the standard DSP probe
panel with 11 additional antibodies which were conjugated to unique
oligos by NanoString's barcoding service [16]. Using standard IHC, we
confirmed the efficient staining performance of the antibodies after oligo
conjugation (Figure 2A) and sufficient DSP counts above the background
control (Figure 2B). In total, antibody selection, conjugation and vali-
dation took 4 months; hence, thorough advance planning of projects is
required.
ROI selection procedure, UV exposure and oligo collection
The DSP platform provides additional benefit to standard IHC or full-
section multiplex platforms by enabling regional/spatial analysis. Based
on the tissue morphology, ROIs can be selected which vary in size
(10–600
μ
m in diameter) and form (Figure 3). The shape of the ROI can
vary from geometric to a rare cell population level. Additional segmen-
tation within a geometric ROI allows distinction to be made between the
tumor and the tumor micro-environment, or between multiple immune
cell types (Figure 3). However, the selection of single cells is not rec-
ommended for the beta instrument at present due to the low signal:noise
ratio of the DSP probes, requiring at least 10 cells/ROI for sufficient
counts. Selection of ROIs results in guided UV light exposure using two
digital micromirror devices (DMDs) in the instrument. These DMDs are
small mechanical systems that contain an array of steerable reflective
micromirrors (Texas Instruments, Dallas, TX). DMD-directed UV light
illuminates all selected ROIs sequentially, resulting in the release of
indexing oligos solely within the boundaries of the set ROI. To confirm
precise guidance of UV light by the DMDs, we used CAGE-532, a dye that
is initially colorless and non-fluorescent but which releases a highly
fluorescent signal when illuminated with UV light (Figure 4A). Further
validation revealed low counts in a control ‘glass’ROI compared with a
Figure 2. Comparison of staining efficiency before and after oligo conjugation of antibodies. (A) Standard immunohistochemical staining for CD16 and CD39 before
and after oligo conjugation of the antibodies. Rabbit immunoglogulin G (IgG) was used as the background control. Scale bar 100
μ
m. (B) Housekeeping (HK)-
normalized digital spatial profiling counts after oligo conjugation of antibodies from three discrete regions of interest (ROI1–3). Violet bars, rabbit IgG; pink bars
(upper figure), CD16; pink bars (lower figure), CD39.
T.M. Van, C.U. Blank Immuno-Oncology Technology 1 (2019) 11–18
13
neighboring tissue ROI (Figure 4B), and precise protein expression ac-
cording to the segment type (Figure 4C).
In our experience, defining representative ROIs requires strong
guidance from an experienced pathologist to avoid biased analysis,
especially with heterogeneous tissues. A detailed description of the
manner of ROI selection will also be a prerequisite in DSP publi-
cations. Our experiments show high reproducibility between com-
parable sections (Figure 5), but strong variations occur in
comparisons of tumor sections with different immune cell abun-
dance (‘hot’vs ‘cold’tumors) (data not shown). Ideally, spatially
structurally delimited but impure tumor regions should be selected.
We therefore prefer the selection of ROIs according to the tumor
structure (e.g. intratumor, peritumor and pure stromal ROIs). To
focus on immune cell populations, segmentation within these
impure tumor ROIs can be performed. In the current state, the DSP
analysis tools could be improved to incorporate the needs of users
analysing complex tissue samples, especially in high-density regions
(Figure 4D). By selecting distinct ROIs, this platform enables users
to obtain spatial data for protein targets relative to each other, as
well as in relation to tissue architecture and immune infiltration.
However, spatial information between certain ROIs can only be
assumed and cannot be measured, and a tool which incorporates a
distance of spatial information would be valuable.
In addition to area-specific characterization of the tissue, full
tissue section analysis can also be performed on this platform by
selecting adjacent square ROIs throughout the tissue. However, in
our opinion, it is not recommended or feasible to carry out this
time-consuming and costly procedure. Therefore, the DSP platform
is not the preferred tool and we recommend the use of specialized
sequential staining platforms, such as CODEX (Akoya Biosciences,
West Coast, 1505 O'Brien Drive, Suite A-1, Menlo Park, CA, USA) or
InsituPlex (Ultivue Inc., Cambridge, MA) [17,18].
Quantification using NanoString's nCounter instrument and data analysis
For digital quantification, the photocleaved oligos are hybridized to
NanoString barcodes and processed on the nCounter instrument. This
results in digital counts corresponding to the abundance of each targeted
protein within each ROI. These digital counts can be analysed in the
provided web-based software, in which data are associated with the
tissue scan and ROIs. Several implemented functions allow comparison of
the acquired DSP probe counts from different ROIs in (clustered) heat-
maps, boxplots and bar graphs, and perform statistical analysis. Our
experience to date with the software is that comparison of data from two
to three slides is simple. However, experimental groups involving more
than three tissues and three points of comparison (e.g. patients' response,
treatment and dose) require careful annotation to enable clustering of the
data, or bioinformatics support.
Due to the high-dimensional analysis of the DSP platform, technical
and biological variations need to be controlled for during data compar-
ison and interpretation. To address various sources of technical variation,
NanoString implemented several internal control mechanisms to
normalize for variables, including External RNA Control Consortium
(ERCC) controls, housekeeping (HK) proteins and immunoglobulin G
(IgG) controls (Table 2). ERCCs are included as positive and negative
controls for technical variation during hybridization, while HK proteins,
number of nuclei or area size are included to normalize the variation of
cellularity within the ROIs (Table 2). To date, there is no established
guideline for normalization; IO pathologists, researchers and bio-
informaticians need to develop consensus guidelines. We found the best
method for analysis was to normalize to ERCC, scale counts to nuclei
counts, and then normalize to either HK proteins or IgG controls. How-
ever, we have also identified some exceptions that require careful
normalization in order to avoid artificial elevation of counts. For
example, when using a geometric ROI on ‘glass’as a background control
Figure 3. Selection of type of region of interest (ROI). Three types of regions can be selected within the digital spatial profiling platform. Tissue biopsy was stained
with S100B/PMEL17 (green), LDH (red) and CD45 (yellow) visualization markers. (A–C) Geometric ROIs can be selected, ranging from circles (A) to rectangles (B) to
user-defined polygons (C). (D,E) Segmentation within a geometric ROI is generated based on visualization markers, and can currently distinguish between tumor and
stroma (D) or specific cell type populations (E). (F) Single-cell ROIs are generated based on visualization markers which allow the analysis of either one or multiple
single cells within a field of view. Scale bar 100
μ
m.
T.M. Van, C.U. Blank Immuno-Oncology Technology 1 (2019) 11–18
14
Figure 4. Precise ultraviolet (UV) guidance by digital mirror devices. (A) CAGE-dye-stained tonsil tissue with geometric region of interest (ROI) before and after UV
exposure. Scale bar 100
μ
m. (B) Circular ROIs (200-
μ
m diameter) on cell pellet array and ‘glass’and respective protein expression levels after digital spatial profiling
(DSP) analysis. DSP counts are normalized to immunoglobulin G (IgG) controls to correct for noise. Scale bar 100
μ
m. (C) Geometric ROI on colorectal cancer tissue
with internal segmentation for ‘tumor’(red) and ‘stroma’(green). Scale bar 100
μ
m. Heatmap of region-specific nCounter counts normalized to nuclei. (D) Circular
ROI (200-
μ
m diameter) with segmentation on ‘PanCKþCD45- CD3-’(blue), ‘PanCK- CD45þCD3-’(yellow) and ‘PanCK- CD45þCD3þ’(red) cells and respective DSP
counts after normalization to IgG controls. Scale bar 100
μ
m.
T.M. Van, C.U. Blank Immuno-Oncology Technology 1 (2019) 11–18
15
or when comparing tumor and rare cell segmentation with a significant
difference in cell number, data cannot be scaled to nuclei or normalized
to HK proteins. This would result in false-positive counts; instead,
normalization to IgG controls is recommended.
Of note, analysis and interpretation of counted indexing oligos have
to be carried out carefully as a single count does not necessarily reflect
protein expression on a single cell. This is due to potential variation in
antigen density between cells and the fact that cells may bind simulta-
neously to multiple oligo-conjugated antibodies. Heterogeneous oligo-
antibody conjugation is regulated and controlled by NanoString to a
certain extent by analysis of the ‘degree of labeling’. Using the nCounter
system, they estimate the oligo numbers for each clone, and remaining
unconjugated antibodies are purified via high-performance liquid chro-
matography [16]. Normalization of the counts to cell nuclei will give an
indication of the density of positive cells or protein expression. However,
these variabilities hinder estimation of the percentage of cells
co-expressing certain markers, unlike other tools such as flow cytometry.
Figure 5. Reproducibility of digital spatial profiling analysis. Independent expression analyses of depicted proteins were performed on two serial sections of (A) cell
pellet array and (B) colorectal cancer tissue (B). Expression levels are shown for one 200-
μ
m-diameter circular region of interest (ROI)/section and counts were
normalized to nuclei and housekeeping proteins. ROIs of serial sections were chosen in the same tissue area to allow close comparison. r
2
value indicates correlation
between expression profiles from sections 1 and 2. Section 1, dark grey bars; section 2, light grey bars.
Table 2
Data analysis and normalization: (A) Technical and biological sources of varia-
tion during the digital spatial profiling (DSP) working process. To subtract the
variation, NanoString Technologies offers ‘different types’of normalization for
protein DSP; (B) Isotype controls and housekeeping proteins included in the DSP
probe panel
A
Source of variation Type of normalization
Technical
variation
nCounter
quantification
ERCC correction
Biological
variation
ROI size Area normalization
ROI cellularity Reference protein/gene
normalization
Nuclei count normalization
ROI background Three isotype controls
B
Isotype controls Mouse IgG1, IgG2a and rabbit IgG
Housekeeping proteins Histone H3, GAPDH and ribosomal protein S6
ROI, region of interest; ERCC, External RNA Control Consortium.
T.M. Van, C.U. Blank Immuno-Oncology Technology 1 (2019) 11–18
16
At present, co-expression of proteins can be identified only for a cell
population within an ROI, but not at single cell level due to the limitation
of single cell analysis. Taking these variabilities into account, we
currently validate our top protein candidates derived from DSP analyses
by standard IHC methods. Our comparisons of DSP counts with staining
obtained using VMs (Figure 6A) or standard IHC reveal good correlation,
enhancing the reliability of the DSP platform (Figure 6B and C).
Conclusion and future perspectives
NanoString's DSP platform is an innovative technology that combines
imaging and tissue sampling to advance the standard IHC procedures
currently used by pathologists. In our view, the major advantages of the
platform are its high multiplexing ability on FFPE samples, requiring low
hands-on time, and the non-destructive straightforward procedure that
will contribute to define biomarker signatures in discrete ROIs. More-
over, the segmentation tool enables users to focus the analysis on, for
example, distinct immune cell infiltrates within a tumor area. As a beta
test site, we have encountered a few correctable instabilities in the
software but not in the hardware of the DSP instrument. Nonetheless,
during the testing phase, we have encountered restrictions in the
simultaneous combination of more than four VMs, which requires the use
of serial tumor tissue sections for different research questions. During this
testing period, single cell analyses were not feasible, which limited our
analytical approaches.
The DSP platform is a rapidly evolving platform that, in the future,
will likely include more DSP probes to cover other aspects of tissue
profiling. In addition to the protein DSP, in the near future, NanoString
will also provide the possibility to characterize RNA expression on FFPE
tissues [14]. However, to implement the DSP platform for translational
research, established guidelines are needed for data analysis, including
normalization strategies, and a more comprehensive understanding of
the effects of staining intensity and oligo labelling on DSP counts. This
will likely be addressed in the near future with the upcoming possibility
Figure 6. Correlation of digital spatial profiling (DSP) counts with immunofluorescent or standard immunohistochemical staining. (A) Tonsil tissue stained with
PanCK (green), CD45 (red) and CD3 (yellow) visualization markers. Scale bar 100
μ
m. Graphs depict nuclei and housekeeping (HK)-normalized DSP counts of
indicated proteins in ‘tumor’or ‘immune’-enriched regions of interest (ROIs). (B) Nuclei and HK-normalized counts for CD163 and PD-L1 obtained from melanoma
patient groups 1 (n¼22) and 2 (n¼32). Each dot represents one ROI/patient selected by similar tissue morphology. *P<0.005, unpaired t-test. ns, non-significant. (C)
Representative images for CD163 and PD-L1 were obtained from melanoma tissue samples from groups 1 and 2 using standard immunohistochemistry. Scale bar
200
μ
m.
T.M. Van, C.U. Blank Immuno-Oncology Technology 1 (2019) 11–18
17
to characterize a single cell. Similar to the recent publication by Decalf et
al. [19], methodological cross-laboratory comparisons between the DSP
platform and other multiplex staining methods, such as sequential
chromogenic IHC multiplexing, Akoya's Codex and Vectra systems, are
essential to optimize the use of each technology based on the research
question and/or clinical need.
In conclusion, in our view, the DSP platform will be an important
addition to current single staining IHC methods in clinical diagnostics.
Recent published data, and our unpublished data, reveal sufficient
characterization of the melanoma protein profile in patient cohorts that
received immune checkpoint blockade therapy using protein DSP [20,
21]. We envisage that with further standardization and optimization of
the working process, data analysis and extension of the visualization and
DSP probe panels, DSP-based multiplex analyses can become a helpful
tool towards personalized immunotherapies.
Funding
CB and TMV received a research grant from NanoString for testing the
beta version of the DSP machine.
Disclosure
CB and TMV received a research grant from NanoString for testing the
beta version of the DSP machine.
Acknowledgements
TMV thanks NanoString Technologies for the opportunity to join the
DSP development team for a 6-month training rotation. We thank Dr AM
Terry for proofreading the manuscript.
References
[1] Chen DS, Mellman I. Oncology meets immunology: the cancer-immunity cycle.
Immunity 2013;39:1–10.
[2] Morisaki T, Katano M, Ikubo A, Anan K, Nakamura M, Nakamura K, et al.
Immunosuppressive cytokines (IL-10, TGF-beta) genes expression in human gastric
carcinoma tissues. J Surg Oncol 1996;63:234–9.
[3] Fleming V, Hu X, Weber R, Nagibin V, Groth C, Altevogt P, et al. Targeting myeloid-
derived suppressor cells to bypass tumor-induced immunosuppression. Front
Immunol 2018;9:398.
[4] Blank CU, Haanen JB, Ribas A, Schumacher TN. Cancer immunology. The "cancer
immunogram". Science 2016;352:658–60.
[5] Chen DS, Mellman I. Elements of cancer immunity and the cancer-immune set
point. Nature 2017;541:321–30.
[6] Baras AS, Drake C, Liu JJ, Gandhi N, Kates M, Hoque MO, et al. The ratio of CD8 to
Treg tumor-infiltrating lymphocytes is associated with response to cisplatin-based
neoadjuvant chemotherapy in patients with muscle invasive urothelial carcinoma of
the bladder. Oncoimmunology 2016;5:e1134412.
[7] Preston CC, Maurer MJ, Oberg AL, Visscher DW, Kalli KR, Hartmann LC, et al. The
ratios of CD8þT cells to CD4þCD25þFOXP3þand FOXP3- T cells correlate with
poor clinical outcome in human serous ovarian cancer. PLoS One 2013;8:e80063.
[8] Shang B, Liu Y, Jiang SJ, Liu Y. Prognostic value of tumor-infiltrating FoxP3þ
regulatory T cells in cancers: a systematic review and meta-analysis. Sci Rep 2015;5:
15179.
[9] Gajewski TF. The expanding universe of regulatory T cell subsets in cancer.
Immunity 2007;27:185–7.
[10] Togashi Y, Shitara K, Nishikawa H. Regulatory T cells in cancer immunosuppression
–implications for anticancer therapy. Nat Rev Clin Oncol 2019;16:356–71.
[11] Blackburn SD, Shin H, Haining WN, Zou T, Workman CJ, Polley A, et al.
Coregulation of CD8þT cell exhaustion by multiple inhibitory receptors during
chronic viral infection. Nat Immunol 2009;10:29–37.
[12] Das M, Zhu C, Kuchroo VK. Tim-3 and its role in regulating anti-tumor immunity.
Immunol Rev 2017;276:97–111.
[13] McGranahan N, Furness AJ, Rosenthal R, Ramskov S, Lyngaa R, Saini SK, et al.
Clonal neoantigens elicit T cell immunoreactivity and sensitivity to immune
checkpoint blockade. Science 2016;351:1463–9.
[14] Merrit CR, Ong GT, Church S, Barker K, Geiss G, Hoang M, et al. High multiplex,
digital spatial profiling of proteins and RNA in fixed tissue using genomic detection
methods. bioRxiv 2019. https://doi.org/10.1101/559021 [epub ahead of print].
[15] Agasti SS, Liong M, Peterson VM, Lee H, Weissleder R. Photocleavable DNA
barcode-antibody conjugates allow sensitive and multiplexed protein analysis in
single cells. J Am Chem Soc 2012;134:18499–502.
[16] Lee J, Geiss GK, Demirkan G, Vellano CP, Filanoski B, Lu Y, et al. Implementation of
a multiplex and quantitative proteomics platform for assessing protein lysates using
DNA-barcoded antibodies. Mol Cell Proteomics 2018;17:1245–58.
[17] Goltsev Y, Samusik N, Kennedy-Darling J, Bhate S, Hale M, Vazquez G, et al. Deep
profiling of mouse splenic architecture with CODEX multiplexed imaging. Cell
2018;174:968–81.
[18] Jungmann R, Avenda~
no MS, Woehrstein JB, Dai M, Shih WM, Yin P. Multiplexed
3D cellular super-resolution imaging with DNA-PAINT and Exchange-PAINT. Nat
Methods 2014;11:313–8.
[19] Decalf J, Albert ML, Ziai J. New tools for pathology: a user's review of a highly
multiplexed method for in situ analysis of protein and RNA expression in tis sue.
J Pathol 2019;247:650–61.
[20] Amaria RN, Reddy SM, Tawbi HA, Davies MA, Ross MI, Glitza IC, et al. Neoadjuvant
immune checkpoint blockade in high-risk resectable melanoma. Nat Med 2018;24:
1649–54.
[21] Blank CU, Rozeman EA, Fanchi LF, Sikorska K, van de Wiel B, Kvistborg P, et al.
Neoadjuvant versus adjuvant ipilimumab plus nivolumab in macroscopic stage III
melanoma. Nat Med 2018;24:1655–61.
T.M. Van, C.U. Blank Immuno-Oncology Technology 1 (2019) 11–18
18
