The shift in cancer therapy from broadly cytotoxic agents toward “personalized” treatments that target specific alterations in each patient's tumor requires diagnostic pathology approaches that are quantitative and biospecimen-friendly. Novel multiplexed antibody-based imaging technologies can measure single-cell expression of over 60 proteins in intact tumor sections and hold promise for clinical oncology.

Since the mid-19th century, the diagnosis and classification of cancer have relied on the description of abnormal cell morphology and tissue structure during review of hematoxylin and eosin (H&E)–stained tumor tissue sections under the light microscope (1). Ancillary molecular tests, in particular targeted DNA resequencing and IHC, have become more critical in recent years for the identification of specific tumor subtypes. IHC testing uses antibodies to determine the expression of a protein target in a formalin-fixed, paraffin-embedded (FFPE) tissue section. Numerous IHC stains are often required to establish an accurate diagnosis and document the expression of proteins that inform the patient's therapy, such as the estrogen receptor (ER) in breast cancer, V600E-mutant BRAF in melanoma, or programmed cell death ligand 1 (PD-L1) in non–small cell lung cancer (NSCLC). The demand for additional tissue sections often exceeds the amount of available tumor, particularly for small tumor biopsies. This pressure point is likely to intensify as proteolysis targeting chimera (PROTAC), antibody–drug conjugates, and targeted radionuclides are unlocking new protein targets for oncology drug development. “Multiplexed” antibody-based imaging technologies can measure the expression of multiple protein targets in a single tumor section and could significantly mitigate this problem (Fig. 1).

Figure 1.

Integration of spatially resolved multiplexed protein profiling into diagnostic pathology. Some images (microscope and computer) are used under license from stock.adobe.com.

Figure 1.

Integration of spatially resolved multiplexed protein profiling into diagnostic pathology. Some images (microscope and computer) are used under license from stock.adobe.com.

Close modal

A second driving force for the development of multiplexed antibody-based imaging assays is the recent discovery in most human cancers of a multitude of coexistent tumor and immune cell functional states, also referred to as “cell phenotypes.” These phenotypes and their transition states were first identified through single-cell RNA sequencing of dissociated tumors (2) and need to be confirmed within the spatial context of the intact tumor microenvironment (TME). It is hoped that a spatially resolved picture of protein expression, posttranslational protein modifications, cell phenotypes, and communities of cell phenotypes will not only advance our understanding of tumor evolution (3) but also guide patient treatment response prediction modeling and the development of novel cancer therapeutics targeting the TME, such as bispecific T-cell engagers, immune-checkpoint inhibitors (ICI), and chimeric antigen receptor (CAR) T cells.

The past 5 years have witnessed an explosion of multiplexed antibody-based imaging technologies that can provide simultaneous single-cell quantification of 2 to 60+ proteins while preserving spatial information, revealing features of cell type (e.g., CD8+ T cell), cell function (e.g., PD1+ CD8+ T cell), and cell localization (e.g., stromal PD1+ CD8+ T cell). These approaches can be applied using only one section of tissue from routinely collected FFPE and/or fresh frozen patient biopsies.

This article provides a brief overview of multiplexed antibody-based imaging technologies, describes examples of their application in cancer research, and discusses challenges toward implementation of these technologies in clinical oncology. Of note, there are multiplexed technologies that are not imaging-based, capture RNA and protein (e.g., NanoString GeoMx Digital Spatial Profiling; ref. 4) or RNA only (e.g., 10X Genomics Visium Spatial Transcriptomics; ref. 5), and can provide complementary views. However, the focus of this review will be on multiplexed antibody-based imaging.

The shared feature between all techniques is the use of protein-specific antibodies to detect multiple proteins of interest in a single assay. The primary antibody bound to a protein epitope can be recognized after direct conjugation with a chromogen, fluorophore, or elemental isotope and subsequently detected using chromogenic, immunofluorescent, or mass spectrometry assays, respectively, or by using a secondary antibody to recognize the primary antibody detected in a similar manner. Staining of antibodies is performed either by adding all antibodies to the slide simultaneously or by using a cyclical and/or sequential stain and strip approach (6). Similarly, imaging of the stain is either performed in a one-step or multistep fashion. The maximal tumor area that can be analyzed is somewhat dependent on the selected method and can range from a single core within a tissue microarray (TMA; millimeters2, typically hundreds to thousands of cells) to the entire tissue section (centimeters2, typically thousands to millions of cells). Here, we will provide an overview of various multiplexed antibody-based imaging technologies (Supplementary Table S1).

Among techniques that use chromogenic detection, multiplexed IHC consecutive staining on single slide (MICSSS) directly builds on IHC. MICSSS uses iterative rounds of staining and chromogen stripping with multistep whole slide imaging. Although MICSSS is a slower assay to complete, it has the advantage of capturing the entire tissue using a standard brightfield scanner without confounding tissue autofluorescence (AF). Additionally, given the widespread use of IHC in the clinical setting, MICSSS has the advantage of generating data most comparable with what pathologists examine in routine diagnostic practice.

A growing number of technologies use immunofluorescence (IF) to quantify the expression of multiple proteins in a single tissue section. Cell DIVE, formerly known as MultiOmyx, uses iterative rounds of staining and fluorophore bleaching with multistep whole slide imaging to achieve up to 60 proteins on a single slide. Cell DIVE takes advantage of a background fluorescence image that is captured following each cycle of staining for AF removal, resulting in low AF images. However, this additional step introduces additional imaging time. PhenoImager HT, formerly known as Vectra Polaris, is a high-throughput yet lower dimensional technology, utilizing multistep staining with primary and secondary antibodies with fluorescent tyramide signal amplification to achieve up to nine proteins on a single slide with one-step whole slide imaging. The PhenoImager HT process requires spectral unmixing to deconvolute overlapping fluorescent spectra. CyCIF (cyclic immunofluorescence) is a method that uses cyclical staining/fluorophore bleaching/imaging (like Cell DIVE), uses open-source software that is compatible with many microscopes, and has been reported to measure up to 60 proteins on a single slide with whole slide imaging. Additional IF-based multiplexed imaging technologies have emerged more recently, including PhenoCycler-Fusion (formerly known as CODEX), iterative indirect immunofluorescence imaging (4i), Orion, COMET, iterative bleaching extends multiplexity (IBEX), and InSituPlex (ISP). PhenoCycler-Fusion uses one-step staining with multistep whole slide imaging to achieve up to 60 proteins. Orion uses not only one-step staining but also one-step whole slide imaging to achieve up to 21 proteins. 4i, IBEX, ISP, and COMET all use multistep staining and imaging to achieve up to 40, 65, 12, and 40 proteins, respectively. Although IBEX and ISP can capture the entire slide, 4i and COMET are currently limited to smaller regions. The number of protein markers that can be measured for many of these higher plex methods is limited by reagent availability or maintenance of tissue or epitope integrity throughout the process.

The last group of multiplexed antibody-based imaging technologies utilizes mass spectrometry for the detection of multiple proteins in a single slide. These technologies produce high-dimensional data (currently up to 40 proteins) and have the advantage of a one-step staining and imaging protocol, resulting in less hands-on time and tissue damage. However, these detection techniques are limited by comparatively slower imaging times, and much of their application thus far has been limited to small tumor areas in the context of TMAs. The latter group of technologies includes imaging mass cytometry (IMC), multiplexed ion beam imaging (MIBI), and multiplexed ion beam imaging by time of flight (MIBI-TOF).

The ability to measure the expression of multiple proteins at single-cell resolution in situ, with the added advantage of protein colocalization, quantitative reporting, and spatial information, has enabled the identification of distinct functional states for many types of cells and their localization within anatomically defined tumor areas (e.g., the tumor/stroma barrier) or functionally defined “neighborhoods” or “cell communities.”

Much of the initial work with multiplexed antibody-based imaging focused on breast cancer (Supplementary Table S2). Using a 32-plex antibody panel and IMC, one study described distinct cell populations and cell interactions and linked them to previously classified breast cancer subtypes (Giesen, Supplementary Table S2). Another study described integrated histologic and immunophenotypic features, including colocalization of multiple markers [e.g., ER, progesterone receptor (PR), Ki-67; Angelo, Supplementary Table S2]. In another study, eight distinct tumor cell phenotypes were identified, linked to established intrinsic breast cancer subtypes, and associated with in vivo tumor uptake of the radiotracer 18F-fluorodeoxyglucose (FDG) using antibodies against members of the glycolysis, hypoxia, and PI3K pathways (Sood, Supplementary Table S2). The expression of 36 proteins in triple-negative breast cancer was measured using MIBI-TOF, revealing archetypes of tumor–immune composition (cold, mixed, compartmentalized) and expression of immunoregulatory proteins in a cell-type and location-specific manner (Keren 2018, Supplementary Table S2). These findings were further extended in a follow-up study (Keren 2019, Supplementary Table S2). A panel of 35 antibodies and IMC were used to define 59 distinct tumor cell phenotypes and distinct microenvironment communities that contained tumor and stromal cell components (Jackson, Supplementary Table S2). In a follow-up study, the authors linked cell phenotypes (including epithelial, stromal, and immune cells) and neighborhoods with comprehensive genomic annotation of the same tumors (Raza Ali, Supplementary Table S2). The inclusion of antibodies against proteins involved in vascular and stromal heterogeneity allowed authors of one study to describe 10 multicellular TME structures that were differentially enriched in distinct breast cancer types and associated with clinical outcomes (Danenberg, Supplementary Table S2). Another study delineated complex cell phenotypes and TME states of normal breast, ductal carcinoma in situ, and invasive breast cancer and identified states that were associated with tumor recurrence (Risom, Supplementary Table S2).

Studies in colorectal cancer build on the earlier finding that not only the type and density of T cells but also their intratumoral localization provides prognostic information (ref. 7; Supplementary Table S2). One study examined 61 protein antigens in 747 colorectal tumor samples and described patterns of coexpression for multiple members of the mTOR and MAPK signaling pathways (Gerdes, Supplementary Table S2). CODEX was used to measure the expression of 56 proteins, and the authors defined nine conserved cellular neighborhoods that were associated with outcomes in patients with colorectal cancer (Schürch, Supplementary Table S2). In another study, PhenoImager HT was used to characterize the density and spatial distribution of natural killer cells, natural killer T-like cells, and other complex cellular phenotypes and their relationship to patient outcomes in colorectal cancer (Väyrynen, Supplementary Table S2).

The ability of multiplexed antibody-based imaging to uncover new aspects of cancer biology has motivated similar studies in other cancer types (Supplementary Table S2). Spatial uniform manifold approximation and projection was developed to identify spatial signatures associated with prognosis in advanced melanoma (Giraldo 2021, Supplementary Table S2). Another study used a combination of IMC and RNA transcript analysis to measure the expression of chemokines and their relationship to immune cell phenotypes and tumor infiltration patterns in metastatic melanoma (Hoch, Supplementary Table S2). Authors of another study reported cell phenotypes and their spatial representation in localized prostate cancer (De Vargas Roditi, Supplementary Table S2).

Another important application of multiplexed antibody-based imaging is the generation of drug response biomarker hypotheses that could inform the clinical development and optimal use of novel cancer therapies. Early tissue-based studies, which did not yet leverage the full potential of these technologies, identified candidate predictive tissue-based biomarkers and documented TME changes in cancer patients treated with ICIs (8–11). These types of studies now increasingly use the latest multiplex antibody-based imaging technologies (Supplementary Table S2).

Several studies have examined tumor tissue before or during the treatment of cancer patients with antibodies against programmed cell death 1 (PD-1), PD-L1, or cytotoxic T-lymphocyte associated protein 4 (CTLA-4). In one instance, pretreatment biopsies from melanoma patients treated with PD-1 blockade were profiled and the expression of 25 tumor- and immune-related protein markers were reported and associated with progression-free survival and overall survival (Martinez-Morilla, Supplementary Table S2). Another study examined the expression of six markers (PD-1, PD-L1, CD8, FOXP3, CD163, and SOX10/S100) in pretreatment tumor biopsies from patients with melanoma receiving anti–PD-1 therapy and reported cell phenotypes associated with treatment response (Berry, Supplementary Table S2). In a different setting, pretreatment tumor biopsies from melanoma patients receiving ICIs were examined for the expression of 35 protein markers. It was found that the presence of proliferating “antigen-experienced” cytotoxic T cells (CD8+CD45RO+Ki-67+) in close proximity to tumor cells was predictive of response to ICIs (Moldoveanu, Supplementary Table S2). Another study examined the composition and spatial distribution of tumor-infiltrating subsets in baseline tumor biopsies from patients with NSCLC treated with PD-1 blockade (Lopez de Rodas, Supplementary Table S2). The authors reported that a higher density of stromal CD8+ T cells was significantly associated with improved progression-free and overall survival in patients with PD-L1+ tumors. An evaluation of pretreatment tumor samples from patients with Merkel cell carcinoma receiving anti–PD-1 checkpoint blockade revealed that the presence or absence of PD-L1 expression alone was not predictive of response, whereas the density of PD1+ and PD-L1+ cells and the frequency of PD1+ cells in close proximity to PD-L1+ cells correlated with clinical response (Giraldo 2018, Supplementary Table S2). Interestingly, a meta-analysis of studies that assessed the prediction of response to PD-1/PD-L1–directed therapy for cancer suggested that multiplexed antibody-based imaging and multimodal biomarker strategies are superior in predicting treatment response compared with PD-L1 IHC, tumor mutation burden, or gene expression profiling alone (12).

Other studies have examined tumor tissue before or during treatment with other immune-directed therapies. For example, one study examined baseline and on-treatment tumor biopsies from patients with advanced melanoma receiving intratumoral injections of a modified herpes simplex virus type 1 (talimogene laherparepvec, TVEC) and systemic administration of the anti–PD-1 antibody pembrolizumab (Ribas, Supplementary Table S2). The authors reported broad changes in immune cell infiltration and increased expression of PD-L1 following injection of TVEC. In a study of matched untreated and IL2-injected “in-transit” melanoma metastases, changes in cell phenotypes following IL2 administration and predictive biomarkers of IL2 response were examined (Pourmaleki, Supplementary Table S2). Regressed lesions were characterized by nonproliferating CD8+ T cells lacking expression of PD-1, LAG3, and TIM3. In contrast, pretreatment lesions from patients who showed a complete IL2 response were characterized by proliferating CD8+ T cells with an exhausted phenotype (PD-1+LAG3+TIM3+), stromal B-cell aggregates, and membranous tumor cell expression of MHC-I.

Given the remarkable progress with spatially resolved multiplexed protein profiling technologies, the current pipeline of cancer therapeutics targeting novel proteins existing in complex spatial topologies, and the increasing reliance on molecular data for the accurate diagnosis of cancer, it seems timely to ask what it will take to bring these multiplexed antibody-based imaging technologies to clinical pathology and patients with cancer.

First, further progress needs to be made toward harmonizing the reporting of multiplexed antibody-based imaging data (13). It is often difficult to compare findings from different studies, even studies reporting on the same cancer type using the same methodology. This is due to differences in reagents (i.e., antibody clones and antibody conjugates) and their validation, cell segmentation, managing signal and background, cell type and phenotype assignment, and spatial metrics in addition to the amount and type of publicly accessible source data and often limited clinical annotation of patient samples. Although data reporting must be quantitative, pathologists should be able to easily access the raw data (i.e., images), requiring the need for a standard digital image viewer.

Second, there is a need to rigorously test the performance characteristic of each method (e.g., linearity of the assay, sensitivity, specificity, and reproducibility) in clinical samples. These efforts have begun for smaller multiplexed IF panels (14) but must be expanded to larger panels and other technologies. To do so, we must address several questions. How do we distinguish findings that are statistically significant from findings that are not only statistically significant but also biologically meaningful and perhaps even clinically relevant? In planning a study, how much data (e.g., number of patient samples, number of cells, number of proteins) are typically needed to reach certain types of conclusions without overfitting data? And what type of tissue should be included as positive and negative staining controls? Greater priority and resources should be given to prospective Institutional Review Board–approved studies with well-defined biospecimen collection methods, validated antibody panels, and preplanned analysis goals and primary/secondary study endpoints.

Third, early consideration should be given to the logistical aspects of translating cancer discovery science to hospital workflows. Most current studies have been conducted in individual research labs, replete with manual steps and individualized decision algorithms. In contrast, most hospital pathology departments heavily rely on automated staining and data analysis pipelines to allow rapid scaling to meet demand and the tight timelines required for clinical decision-making (days). Furthermore, pathologists rapidly scan through tissue sections from multiple tissue blocks to arrive at the correct diagnosis, whereas many multiplexed antibody-based imaging technologies are limited to the analysis of very small preselected areas within each cancer specimen. Technologies that allow for whole slide scanning, ideally from multiple different tumor blocks for each patient, would mirror current hospital practices for histopathologic assessment and may provide a more accurate representation of the intratumoral heterogeneity found in many human cancers.

Lastly, it will be important to develop a financial framework that will consider how multiplexed antibody-based imaging diagnostics, perhaps assembled into disease-specific panels, could gradually replace time-honored but biospecimen-intensive and largely “semiquantitative” methods of protein detection. It is also important to consider how multiplexed antibody-based imaging could be efficiently integrated with the genomic annotation of cancer samples to achieve integrated diagnostics, accelerate clinical drug development, and ultimately promote cancer care that is more effective, more affordable, more accessible, and less toxic.

N.D. Socci reports grants from the NIH during the conduct of the study. T.J. Hollmann reports grants from Calico Labs and Bristol Myers Squibb, and other support from Ultivue outside the submitted work. I.K. Mellinghoff reports personal fees from Roche Therapeutics, Servier Pharmaceuticals, Black Diamond Therapeutics, Prelude Therapeutics, Novartis, and Voyager Therapeutics, and grants from Puma Biotechnology and General Electric outside the submitted work. No disclosures were reported by the other author.

This study was supported by NIH grant 1 F31 CA271778 01 (M. Pourmaleki), NIH grant 1 R35 NS105109 03 (I.K. Mellinghoff), the Geoffrey Beene Cancer Research Center (I.K. Mellinghoff), Cycle for Survival (I.K. Mellinghoff), and NIH grant 1 P30 CA008748.

Note: Supplementary data for this article are available at Cancer Discovery Online (http://cancerdiscovery.aacrjournals.org/).

1.
van den Tweel
JG
,
Taylor
CR
.
A brief history of pathology: preface to a forthcoming series that highlights milestones in the evolution of pathology as a discipline
.
Virchows Arch
2010
;
457
:
3
10
.
2.
Kinker
GS
,
Greenwald
AC
,
Tal
R
,
Orlova
Z
,
Cuoco
MS
,
McFarland
JM
, et al
.
Pan-cancer single-cell RNA-seq identifies recurring programs of cellular heterogeneity
.
Nat Genet
2020
;
52
:
1208
18
.
3.
Rozenblatt-Rosen
O
,
Regev
A
,
Oberdoerffer
P
,
Nawy
T
,
Hupalowska
A
,
Rood
JE
, et al
.
The Human Tumor Atlas Network: charting tumor transitions across space and time at single-cell resolution
.
Cell
2020
;
181
:
236
49
.
4.
Merritt
CR
,
Ong
GT
,
Church
SE
,
Barker
K
,
Danaher
P
,
Geiss
G
, et al
.
Multiplex digital spatial profiling of proteins and RNA in fixed tissue
.
Nat Biotechnol
2020
;
38
:
586
99
.
5.
Ståhl
PL
,
Salmén
F
,
Vickovic
S
,
Lundmark
A
,
Navarro
JF
,
Magnusson
J
, et al
.
Visualization and analysis of gene expression in tissue sections by spatial transcriptomics
.
Science
2016
;
353
:
78
82
.
6.
Hickey
JW
,
Neumann
EK
,
Radtke
AJ
,
Camarillo
JM
,
Beuschel
RT
,
Albanese
A
, et al
.
Spatial mapping of protein composition and tissue organization: a primer for multiplexed antibody-based imaging
.
Nat Methods
2022
;
19
:
284
95
.
7.
Galon
J
,
Costes
A
,
Sanchez-Cabo
F
,
Kirilovsky
A
,
Mlecnik
B
,
Lagorce-Pagès
C
, et al
.
Type, density, and location of immune cells within human colorectal tumors predict clinical outcome
.
Science
2006
;
313
:
1960
4
.
8.
Tumeh
PC
,
Harview
CL
,
Yearley
JH
,
Shintaku
IP
,
Taylor
EJM
,
Robert
L
, et al
.
PD-1 blockade induces responses by inhibiting adaptive immune resistance
.
Nature
2014
;
515
:
568
71
.
9.
Herbst
RS
,
Soria
JC
,
Kowanetz
M
,
Fine
GD
,
Hamid
O
,
Gordon
MS
, et al
.
Predictive correlates of response to the anti-PD-L1 antibody MPDL3280A in cancer patients
.
Nature
2014
;
515
:
563
7
.
10.
Chen
PL
,
Roh
W
,
Reuben
A
,
Cooper
ZA
,
Spencer
CN
,
Prieto
PA
, et al
.
Analysis of immune signatures in longitudinal tumor samples yields insight into biomarkers of response and mechanisms of resistance to immune checkpoint blockade
.
Cancer Discov
2016
;
6
:
827
37
.
11.
Forde
PM
,
Chaft
JE
,
Smith
KN
,
Anagnostou
V
,
Cottrell
TR
,
Hellmann
MD
, et al
.
Neoadjuvant PD-1 blockade in resectable lung cancer
.
N Engl J Med
2018
;
378
:
1976
86
.
12.
Lu
S
,
Stein
JE
,
Rimm
DL
,
Wang
DW
,
Bell
JM
,
Johnson
DB
, et al
.
Comparison of biomarker modalities for predicting response to PD-1/PD-L1 checkpoint blockade: a systematic review and meta-analysis
.
JAMA Oncol
2019
;
5
:
1195
204
.
13.
Schapiro
D
,
Yapp
C
,
Sokolov
A
,
Reynolds
SM
,
Chen
YA
,
Sudar
D
, et al
.
MITI minimum information guidelines for highly multiplexed tissue images
.
Nat Methods
2022
;
19
:
262
7
.
14.
Taube
JM
,
Roman
K
,
Engle
EL
,
Wang
C
,
Ballesteros-Merino
C
,
Jensen
SM
, et al
.
Multi-institutional TSA-amplified multiplexed immunofluorescence reproducibility evaluation (MITRE) study
.
J Immunother Cancer
2021
;
9
:
e002197
.
This open access article is distributed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0) license.