Multiplex immunofluorescence staining and image analysis assay for diffuse large B cell lymphoma

Journal Pre-proof Multiplex immunofluorescence staining and image analysis assay for diffuse large B cell lymphoma

Chung-Wein Lee, Yan J. Ren, Mathieu Marella, Maria Wang, James Hartke, Suzana S. Couto PII:

S0022-1759(19)30287-X

DOI:

https://doi.org/10.1016/j.jim.2019.112714

Reference:

JIM 112714

To appear in:

Journal of Immunological Methods

Received date:

22 July 2019

Revised date:

23 October 2019

Accepted date:

25 November 2019

Please cite this article as: C.-W. Lee, Y.J. Ren, M. Marella, et al., Multiplex immunofluorescence staining and image analysis assay for diffuse large B cell lymphoma, Journal of Immunological Methods (2019), https://doi.org/10.1016/j.jim.2019.112714

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

© 2019 Published by Elsevier.

Journal Pre-proof Title Multiplex Immunofluorescence Staining and Image Analysis Assay for Diffuse Large B Cell Lymphoma

Chung-Wein Lee* , Yan J. Rena, Mathieu Marellaa, Maria Wangb, James Hartkeb, Suzana S.

ro

of

Coutob

-p

a: Department of Non-clinical Drug Safety, Celgene Corporation, 10300 Campus Point Dr. San

re

Diego CA, 92121 USA (These authors contributed equally to this work)

lP

b: Department of Non-clinical Drug Safety, Celgene Corporation, 10300 Campus Point Dr. San

na

Diego CA, 92121 USA

*: Corresponding author at: Department of Non-clinical Drug Safety, Celgene Corporation,

Jo

ur

10300 Campus Point Dr. San Diego CA, 92121 USA, Email address: [email protected]

Summary of Declarations of Interest: None.

Journal Pre-proof Abstract With the explosion of immuno-oncology and the approval of many immune checkpoint therapies by regulatory agencies in the last few years, understanding the tumor microenvironment (TME) in the context of patient’s immune status has become essential. Among available immune profiling techniques multiplex immunofluorescence (mIF) assays offer the unique advantage of preserving the architectural features of the tumor and revealing the spatial relationships between tumor cells and immune cells. A number of mIF and image analysis assays have been described for solid tumors but most are not sufficiently suitable for lymphoma, in which the lack of clear

of

tumor-stromal boundaries, and high tumor density present significant challenges. Here we

ro

describe the development and optimization of a reliable workflow using Akoya Opal staining kit to label and analyze 6 markers per slide in diffuse large B-cell lymphoma (DLBCL). Five panels

-p

of 30 markers were developed to characterize infiltrating immune cells and relevant check-point

re

proteins such as PD1, PDL1, ICOS, SIRP-alpha and Lag3 on 70 DLBCL sections. Multiplexed sections were scanned using Akoya multispectral scanner. Image analysis workflow using

lP

InForm and Matlab was developed to overcome challenges inherent to the DLBCL environment. Using the assays and workflows detailed here we were able to quantify cell densities of subsets

na

of infiltrating immune cells and observe their spatial patterns within the tumors. We highlight heterogeneous distribution of cytotoxic T cells across tumors with similar T cell density to

ur

underscores the importance of considering spatial context when studying the effects of

Jo

immunological therapies in DLBCL.

Keywords: Immunophenotyping; Lymphoma; Multiplex immunofluorescence; Multispectral imaging; Image analysis; Digital pathology.

Journal Pre-proof 1. Introduction With the approval of numerous immunotherapies in the past few years, understanding the tumor microenvironment has become essential to contextualize clinical response, inform combinations and select appropriate indications. Due to the complexity and dynamic nature of the tumor immune microenvironment, there is a need for more advanced techniques to capture specific subsets of immune cells and look at cell-cell relationships in patient samples. The PD-1/PD-L1 axis blockade offers a good example of this complexity. In spite of the success of the checkpoint inhibitors for PD-1 (nivolumab and pembrolizumab) and PD-L1 (atezolizumab and durvalumab)

of

in solid tumor indications and melanoma, resistance and relapse were also observed in spite of

ro

PD-L1 expression in the tumors (Hamid et al., 2013; Topalian et al., 2012). More recently, combined analysis of CD8+ cytotoxic cells and PD-L1 expression in tumors has been described

-p

as a more reliable prognostic marker to PD-L1 alone in gastric cancer (Morihiro et al., 2019).

re

Further understanding of the immunological responses to drugs in the tumor microenvironment and development of more sophisticated prognostic biomarkers are key to design treatment

lP

regimens tailored for individual patients (Binnewies et al., 2018). Currently, the most commonly used tissue-based techniques in immune oncology research

na

include RNA sequencing, flow cytometry and multiplex immunofluorescence (mIF). RNA sequencing is a high-throughput technique and has been used to generate immune signatures and

ur

prognostic classifiers for multiple tumor types (Binnewies et al., 2018; C-W Lee, 2010;

Jo

Rosenwald et al., 2002). However, low RNA quality of formalin- fixed, paraffin-embedded tissues limits its application in archival or diagnostic samples. Additionally, tissue architecture is destroyed in sample preparation resulting in loss of spatial context. Flow cytometry is multichannel, sensitive, and able to quickly identify different cell types by size and markers and has been used for immunophenotyping of peripheral blood and tumor samples (Finak et al., 2016; Metrock et al., 2017). However, it requires fresh samples and dissociation of cells which, similarly to RNAseq, leads to loss of tissue architecture and relevant spatial relationships between immune cells and tumor cells (Dunphy, 2004). mIF, on the other hand, can display expression of multiple markers in individual cells while preserving spatial relationships between immune cells and tumor cells, and has become a powerful tool for characterization of the tumor immune microenvironment (Tsujikawa et al., 2017). However, the usefulness of the panels and

Journal Pre-proof staining quality can be affected by many factors, such as limited number of markers per panel, steric hindrance of the antibody to epitopes, auto-fluorescence, leaking of light between different fluorophore channels and challenges pertaining to scanning images and analyzing the results (Dixon et al., 2015; Gorris et al., 2018). It is thus essential to carefully design and optimize each multiplex panel, not only for the selected markers, but also in regard to the fluorophores and detection system and appropriate image capture and analysis. Once a high quality image is obtained, an equally important step includes the creation of accurate cellular segmentation and phenotyping algorithms and a workflow that can be verified for accuracy. In multiplexed

of

fluorescence image acquisition, the contamination of auto-fluorescence and the fluorescence

ro

bleed-through affect the quality of image visualization and the detection of low level marker expression because of inadequate signal to noise ratios. Here we have opted to use the Akoya

-p

Bio Opal Fluorophore detection system and Akoya Vectra 3 multispectral scanner. The Opal system uses a tyramide based amplification (TSA) to detect low markers and the Vectra 3

re

scanner can accurately separate the convoluted fluorescence image into the components from

lP

each fluorescence labeled antibodies using multispectral technology (Parra et al., 2017; Stack et al., 2014). For mIF image quantification, a semi-automated image analysis workflow combining conventional image analysis algorithms and machine learning is beneficial, especially when the

na

number of images is large. Recently, rapid progress in computation hardware and artificial intelligence algorithms mitigate the difference between the scoring done by the pathologist and

ur

the computer-aid approach. However pathologist review on computer-aid analysis results is still

Jo

important to confirm the accuracy of cell detection and correct phenotyping. A number of published reports describe the optimization of mIF for solid tumors, but few are fully automated or reproducible for large number of samples. In this technical report we describe an optimized digital pathology workflow to attenuate the errors in staining, image acquisition and analysis in diffuse large B-Cell lymphoma (DLBCL). We designed five 6-plex panels (30 immune markers in total) to investigate the immune landscape across 70 DLBCL samples. Multispectral images were analyzed using a two-tiered approach to increase accuracy. First, a conventional nuclear segmentation and machine learning algorithm with binary classification scheme were used to detect the cell population expressing only a single marker, and then another set of algorithms were applied to determine other types of immune cells co-expressing multiple markers. The workflow shared here is reproducible and provides accurate cell phenotyping in

Journal Pre-proof challenging lymphoma samples. The spatial distribution of infiltrating immune cells in DLBCL is highly heterogeneous and further characterization of cell-cell interaction may help us predict

Jo

ur

na

lP

re

-p

ro

of

clinical responses and mechanism of resistance to immune-oncology therapies.

Journal Pre-proof 2. Materials and Methods

2.1 Multiplex immunofluorescence (mIF) staining The detection method of the mIF panels described here is based on the use of fluorophores conjugated to a tyramide molecule (PerkinElmer, Waltham, MA; Catalog No. NEL801001KT) and is illustrated in Fig. 1. All steps were performed on the Leica Bond Rx autostainer (Leica Microsystems, Buffalo Grove, Illinois). The main advantage of this detection system is that it does not require specially labeled antibodies and can be used with unmodified primary

of

antibodies previously validated for chromogenic immunohistochemistry (IHC). In this system,

ro

horse-radish peroxidase is used to catalyze a reaction between tyramide and tyrosine residues on or near the epitope of interest and to covalently deposit each fluorophore to the tissue section

-p

(Chao et al., 1996; van Gijlswijk et al., 1997). This step is followed by heat-mediated striping of the primary and secondary antibodies without removing the fluorophore. A new set of antibodies

re

is then applied to detect the next marker, and the process is repeated sequentially until all 6

lP

markers of interest are detected in a single tissue section (Fig. 1A). These successive staining steps prevent the cross-reactivity between antibodies, and allow multiple primary antibodies to be used without concern for the species in which they were raised. The first consideration when

na

designing a panel is that all primary antibodies are amenable to the same antigen retrieval conditions, usually at pH 9.0, and that the staining patterns are as expected based on known

ur

positive controls and strict validation steps (Supplementary montage). The second consideration

Jo

is the staining order and antibody-fluorophore pairing, which is based on 3 main points: (1) the epitope heat resistance relating to the maximum number of heat mediated striping cycles a specific molecular target could withstand (Fig. 1B), (2) the steric hindrance or masking effect resulting in suboptimum detection of markers that share very close subcellular location (Fig. 1C), and finally (3) the abundance of the targets. A very abundant target should be preferentially associated with a fluorophore of a relative weaker intensity. The relative intensity of each fluorophore was detected by Vectra3 imaging platform (Fig. 1D). Details on antibody order, dilution and fluorophore pairing for the 5 panels described here are shown in Table 1. For all multiplex IF panels, 4 µm tissue sections were incubated in ER2 solution (pH 9, Leica Microsystems) for 20 minutes using the default heat settings on Leica Bond RX. Primary antibody number 1 was incubated for 30 minutes, followed by the Opal

Journal Pre-proof polymer system and the selected Opal TSA fluorophore, each incubated for 10 minutes. A heat mediated stripping step with ER1 solution (pH 6, Leica Microsystems) for 20 minutes at 95 degrees Celsius was inserted between each antibody staining round. Staining with the second antibody followed, and the protocol was repeated sequentially until all markers had been stained. Tissue sections were then counterstained with spectral DAPI for 5 min, rinsed in water, and mounted with ProLong Diamond Antifade Mountant (Life Technology, Carlsbad, CA; Catalog

Jo

ur

na

lP

re

-p

ro

of

No. P36961). The detailed multiplex staining procedures are listed in Supplementary Protocol.

lP

re

-p

ro

of

Journal Pre-proof

Fig. 1: Schematic diagram of multiplex IF panel design and optimization. (A) The multiplex

na

fluorescent staining is based on 3 major steps: (1) specific antibody and an HRP conjugated secondary antibody bind to the protein of interest, (2) the tyramide- fluorophore is activated

ur

through reaction with the HRP and forms covalent bonds with the tyrosine residues present in the near vicinity of the protein of interest. Finally, (3) a heating step removes the primary and

Jo

secondary antibodies leaving only the covalently bound fluorophores on the tissue. The process is then repeated with every fluorophore to generate a multiplexed stained tissue section. (B) Epitope stability is evaluated following several cycles for heat-mediated antibody stripping. Example 1 (blue) shows a stable level of fluorescence, and examples 2 and 3 show denaturation of epitope and decreased intensity and increased intensity due to continued epitope retrieval, respectively. (C) Hindrance and masking effects of the Opal dye system can be a problem when targets share the same subcellular location. Hindrance affects antibody binding due to the potential epitope modification by the covalent bond created by the tyramide-fluorophore. Masking effect is the disturbance of the covalent binding of the subsequent activated tyramidefluorophore by an already bound fluorophore. (D) Relative intensity of each Opal fluorophore should be considered when choosing antibody-fluorophore pairs.

Journal Pre-proof Table 1

Antibodies

Dilutions

Opal fluorophores

Opal dilutions

1

FoxP3

1/200

Opal-520

1/100

2

CD3

1/500

Opal-540

1/100

3

CD20

1/1000

Opal-690

1/100

4

CD8

1/400

Opal-570

1/100

5

PD-1

1/200

Opal-620

1/100

6

CD56

1/800

Opal-650

1/100

1

CD163

1/200

Opal-520

of

Order

1/100

CD3

1/500

Opal-540

3

CD20

1/1000

Opal-690

1/100

4

PD-L1

1/100

Opal-570

1/100

5

C-MAF

1/300

Opal-620

1/100

6

CD11c

1/200

Opal-650

1/100

1

CD163

1/200

Opal-520

1/150

2

CD3

1/500

Opal-540

1/150

3

CD20

1/1000

Opal-690

1/150

4

SIRP-alpha

1/600

Opal-570

1/150

5

CD47

1/100

Opal-620

1/150

6

CD11c

1/200

Opal-650

1/150

1

CD3

1/500

Opal-520

1/150

2

FoxP3

1/800

Opal-540

1/150

3

ICOS

1/50

Opal-570

1/150

4

CD20

1/1000

Opal-620

1/150

5

Granzyme-B

1/200

Opal-650

1/150

6

CD8

1/200

Opal-690

1/150

1

CD3

1/500

Opal-520

1/150

2

TIM3

1/200

Opal-540

1/150

3

CD20

1/1000

Opal-570

1/150

4

Lag3

1/400

Opal-620

1/150

5

CD163

1/200

Opal-650

1/150

6

PD-1

1/200

Opal-690

1/150

Jo

ur

na

lP

-p

ro

2

re

Lag3 panel

ICOS panel

SIRPalpha panel

PD-L1 panel

PD-1 panel

Five multiplexed panel information

1/100

Journal Pre-proof 2.2 Multiplexed IF image acquisition and analysis A schematic overview of the workflow from staining to image analysis is shown in Fig. 2. Vectra 3 (Akoya Biosciences Inc., Menlo Park, CA) was first used to scan the entire fluorescence slide (whole slide scan) at 4X magnification. The whole slide image was then loaded into Phenochart (Akoya Biosciences Inc., Menlo Park, CA) for the pathologist to manually annotate all tumor regions of interest (Fig. 2.3). The edges of the tissue were excluded in annotation because they were likely affected by staining artifacts. The annotated regions were then scanned at 20X in multispectral fashion at 20nm wavelength intervals between 420nm and 720nm. Eight

of

DLBCL slides were used to build a standard spectral library, six for each Opal fluorophore; one

ro

for DAPI only; one without any fluorescent staining to quantify the intrinsic autofluorescence of the DLBCL samples. The spectral library slides were also scanned at 20X and the images were

-p

loaded into inForm (Akoya Biosciences Inc., Menlo Park CA) to create the standard spectral

re

library. The library was used to unmix the 7-color DLBCL multiplexed images to create single component images which show the distribution of each antibody in the section (Fig. 2.4).

lP

A two-tiered image analysis workflow was implemented using InForm and Matlab (Mathworks Inc., Natick MA). In the first tier InForm was used for cell segmentation and binary phenotyping.

na

For cell segmentation several low, medium, and high DAPI intensity images were first loaded into inForm. The parameters such as DAPI intensity, minimum nuclear size, and splitting factor

ur

were adjusted to reach good nuclear segmentation. For panels with other nuclear markers, such

segmentation.

Jo

as FoxP3 and CMAF, the additional nuclear markers were also included to assist nuclear

For phenotyping each marker we took a binary classification scheme to classify cells into a positive category and a negative category, similar to the scheme “one-vs-all” adapted in support vector machine (Ryan Rifkin, 2004) (Fig. 2.6). Because there are 6 markers in a panel this binary classification has to be repeated 6 times. The phenotyping algorithm of inForm is based on machine learning and considers the features of both intensity and morphology in classifying cells (Friedman et al., 2010). In binary phenotyping several images with low, medium, and high marker intensity were first loaded into inForm, the representative positive cells, which expressed the marker, and the negative cells, which did not express the marker, were selected as the training set to train the algorithm. The algorithm was then tested on several additional images.

Journal Pre-proof The optimized cell segmentation and phenotyping algorithm were applied to analyze all DLBCL 20X images. The final cell segmentation data and binary phenomaps for each marker were saved for the second tier analysis. In a second tier, the phenomaps from all 20 X images were uploaded into Matlab to calculate the number of all nucleated cells and the number of positive cells of each marker. The dimension of a 20X image is 0.67mm by 0.5mm which is equal to 0.335 mm2 in area. For every 20X image the cell density for each cell type was calculated as the number of positive cells per area (cell counts/mm2 ), or as a ratio of the positive cells over total nucleated cells. The average cell density

of

and standard of deviation (STD) of all 20X images for a given DLBCL sample were calculated

ro

to indicate population and heterogeneity. Immune cells that are positive for more than one marker can be quickly identified by overlapping the corresponding binary phenomaps in Matlab

-p

(Fig. 2.8). Cell density or ratio for immune cell populations positive for more than one marker

re

(e.g. dual positive of CD3+ and Foxp3+) were calculated in the same manner as described above

Jo

ur

na

lP

for single markers.

lP

re

-p

ro

of

Journal Pre-proof

Fig. 2 Schematic overview of multiplexed immunofluorescence (mIF) workflow. (1) DLBCL

na

tissue architecture overview was evaluated under microscope, low tissue quality sections were excluded from multiplex staining. (2) Leica BOND autostainer was used for all multiplexed

ur

staining and standard spectral slide preparation. (3) Once the whole slide scan at 4X was completed, the pathologist outlined the region of interest (ROI) for multispectral imaging at 20X,

Jo

each red box in the grid representing a single 20X image location in the DLBCL section. (4) After multispectral scans, 20X images were unmixed to the component images for image analysis. (5) Before segmenting nuclei in 20X images, the average DAPI intensity was calculated, extremely dim DAPI sections (indicated by a red arrow) were removed or restained with DAPI after consulting with the pathologist. (6) Binary cell phenotyping using machine learning was conducted to create six sets of binary phenomaps, the single positive marker cell densities were calculated using Matlab. (7) Cell density distribution for the single marker was used to evaluate the accuracy of binary phenotyping. The poor binary phenotyping performance were often found at the sections with extremely low or high cell density as indicated by two red arrows. More training cells from those sections were added into machine learning algorithm to improve accuracy of binary phenotyping. The steps 6 to 7 were done iteratively until the

Journal Pre-proof pathologist approves the phenotyping performance. (8) The cells coexpressing multiple antibodies were identified by overlapping the single binary phenomaps. The cell density and the

Jo

ur

na

lP

re

-p

ro

of

location in every 20X images were calculated and recorded using Matlab.

Journal Pre-proof

2.3 Sanity checks implemented along the entire workflow Several sanity checks were implemented throughout the entire workflow. A hematoxylin and Eosin (H&E) slide of each sample was examined by the pathologist to confirm tissue quality, the failed DLBCL sections were excluded from the multiplexed staining (Fig. 2.1). Automatic BOND RX stainer (Leica Microsystems, Buffalo Grove, Illinois) was used for developing multiplex panels and for all batch staining tasks to maintain staining consistency (Fig. 2.2). We

of

used two to three tonsil sections to inspect the overall staining fidelity of all markers for every

ro

batch staining using Leica BOND.

In image analysis workflow nuclear segmentation was the first step to identify cells, the quality

-p

of DAPI stain impacted the outcome of nuclear segmentation. We calculated the average DAPI intensity distribution of all 20X images of the DLBCL section, then excluded poor DAPI stained

re

sections from nuclear segmentation (Fig. 2.5). The performance of phenotyping algorithms using

lP

machine learning algorithm is affected by several factors, such as poor feature selection in training and overfitting (Fiore et al., 2012; Koelzer et al., 2019). In evaluating binary

na

classification, we overlapped the phenomap from the classification algorithm over the marker component image to first visually inspect the algorithm performance. The poor classification

ur

performance, which is easily recognized by the naked eyes, often occurred in the sections with either very low or very high cell intensity (Fig. 2.7). Cells from those sections with poor binary

Jo

classification accuracy were added to the training set to further improve the binary classification. The retrained algorithm was then applied to phenotype all DLBCL sections. The classification performance inspection was conducted again, more additional cells would be added to the binary classification algorithm if the poor classification was observed. The classification inspection continued iteratively (Fig. 2.6 and Fig. 2.7) until the final binary classification outcome was approved by the pathologist.

Journal Pre-proof 3. Results and Discussion 3.1 Multiplex panel staining shows good reproducibility For this reproducibility study, inter- and intra-assay variability was evaluated using two normal lymph nodes (LNa and LNb). Three serial sections were cut from each block and stained on three different days using the PD1 panel protocol. Sections were then scanned using identical settings and phenotyped using the same algorithm. Table 2 shows the coefficient of variance (CV) of cell densities for all 6 markers in the panel which were used to measure the consistency

of

of the staining within the same run (intra-assay), and across separate runs (inter-assay). Factors influencing intra- and inter-assay variability are reagents, autostainer performance, variation of

ro

section thickness, scanner performance, and change of cell composition in the serial sections.

-p

For abundant markers such as CD20 (over 6000 cells/mm2 ), the highest intra-run CV was 6% and the highest inter-run CV was 4%, demonstrating good staining consistency. For less dense

re

markers such as PD-1 (less than 1 cell/mm2 ), CVs were much higher, reaching 47% and 85% for intra- and inter-assay respectively. Higher variability is expected in the case of low density cell

lP

populations, and it is mainly due to change in cell composition in the sample, across serial sections or in different fields of view. Overall, the intra- and inter-assay CVs were as expected

na

and demonstrated that our staining workflow is highly reproducible and performs consistently.

Jo

to complete.

ur

Reproducibility is important when staining large batches of samples that may take several days

Journal Pre-proof Table 2

CD20

CD3

CD56

CD8

FoxP3

PD1

Ave STD CV Ave STD CV

4711 157.87 3% 4433.35 143.51 3%

7706 105.7 1% 7532.52 229.66 3%

262.62 28.02 11% 180.7 34.42 19%

2437.1 225.75 9% 2189.3 273 12%

129.24 20.97 16% 202.37 27.71 14%

0.24 0.08 34% 1.69 0.8 47%

Day3

Ave STD CV

4378.71 209.83 5%

7733.11 649.15 8%

167.13 6.71 4%

1982.31 54.58 3%

Ave STD CV

6350.75 130.63 2%

1465.97 111.44 8%

363.3 196.75 54%

698.39 28.98 4%

41.9 7.77 19%

0.13 0.04 27%

Ave STD CV Ave STD CV

6161.26 58.45 1% 6367.39 362.26 6%

1390.51 28.47 2% 1329.76 40.44 3%

278.7 156.5 56% 255.95 127.86 50%

698.07 11.69 2% 709.24 76.36 11%

50.57 6.86 14% 48.3 5.79 12%

3.16 1 32% 4.18 0.32 8%

ro

-p

re

lP

133.65 2.32 45.77 0.35 34% 15%

CD20

CD3

CD56

CD8

LNa

# of cell /mm2 Ave STD CV

4507.69 178.18 4%

7657.21 108.83 1%

203.48 51.66 25%

2202.9 227.7 10%

155.09 1.42 41.01 1.06 26% 75%

LNb

Inter-

Jo

ur

na

Day 2 Day 2 Day3

LNb

LNa

Day 1

Intra-

of

# of cell /mm2

Day 1

Intra- and Inter-run multiplexed labeling consistency

FoxP3

PD1

Ave STD CV

6293.13 114.5 2%

1395.41 68.24 5%

299.32 56.57 19%

701.9 6.36 1%

46.92 4.49 10%

2.49 2.11 85%

Journal Pre-proof 3.2 The binary phenotyping approach is accurate and versatile in immune cell profiling Consistency and reproducibility are two key criteria in large number image analysis. Close agreement between the CD3 and CD20 component images to the corresponding binary classification phenomaps was shown in Fig. 3A. Because CD3 and CD20 were included in all five panels, we used their cell density to evaluate performance of binary classification and the concordance of the analysis across all five panels. Significant correlation of the average CD3 and CD20 cell density between two panels (Lag3 and PD1 panels, in this example) was shown in Fig. 3B. Similar correlations were observed for all panels (data not shown). These results suggest

of

that the binary classification approach provides consistent immune cell identification and density

ro

quantification. The advantage of this approach over phenotyping multiple cell types at once is that it detects infrequent cells types such as those simultaneously positive for unexpected

re

-p

markers such as CD3 and CD20.

lP

3.3 Spatial context is important in immune-oncology Recent reports show that understanding the spatial arrangement of immune cells and tumor cells

na

is crucial to predict success of certain immunotherapies (Kather et al., 2018; Schwen et al., 2018). Cell count alone might not be able to provide a complete picture of the immune landscape

ur

for some cancers, and an argument for the use of immune cell topography as biomarker in colon cancer has been recently discussed (Morihiro et al., 2019). In the present study set, there were

Jo

distinctive spatial patterns of infiltrating immune cells such as CD8+ T cells, in relationship to CD20+ tumor cells in DLBCL. Samples with very similar CD8 and CD20 cell densities could show radically distinct CD8+ T-cell patterns (Fig. 3C). The distribution of tumor cells was homogeneous in both cases. However, case 375 showed diffusely distributed CD8+ T while case 361 showed distinct clusters of CD8+ T-cells within the tumor. Understanding the implication of these different topographies of infiltrating immune cells in clinical response may shed light in the biology of DLBCL and may help improve the prediction in prognosis and response to immunotherapies.

ur

na

lP

re

-p

ro

of

Journal Pre-proof

Fig. 3 Concordance of analysis and spatial contexture of immune landscape. (A) The left two

Jo

images are the examples of 20X images of CD20 and CD3 colored in cyan together with DAPI component colored in blue. The corresponding phenomaps identified by the binary classification are shown on the right, the positive CD3 or CD20 cells were stamped with red dots, the CD3/CD20 density of the image are respectively 1809.0/mm2 and 8510.4/mm2 . (B) The correlation of CD3 and CD20 cell density between Lag3 panel and PD1 panel of 70 DLBCL samples. Each dot in the plot represents the average CD3 or CD20 cell density of all 20X images of the section. The R and p-value of correlation are shown in the text box at top left corner. (C) Cytotoxic lymphocytes, CD8 positive, are colored in green; tumor cells CD20 in red. DAPI in blue indicates the size of entire tissue, the majority of both DLBCL sections were scanned at 20X to investigate the contexture of immune cells. Both sections (DLBCL375 vs. DLBCL361)

Journal Pre-proof have identical cell density of CD20 (8397/mm2 vs. 8926/mm2 ) and CD8 (1386/mm2 vs. 1455/mm2 ); the CD8 distribution in DLBCL375 is homogeneous, that in DLBCL361 is

Jo

ur

na

lP

re

-p

ro

of

clustered. The yellow scale bar represents 1.0mm in length.

Journal Pre-proof 4. Conclusion Understanding the complexity and intricacy of the tumor immune microenvironment has become essential in immuno-oncology research. Characterization of single markers is slowly being replaced by immune signatures based on mRNA expression data, flow cytometry and multiplex immunofluorescence (mIF). mIF is unique in its ability to provide both expression and location of several immune markers while preserving tissue architecture and spatial context. However, several technical considerations are important in developing a multiplex panel. Detailed quality control of every step, from staining to analysis, is necessary to ensure accurate

of

immunophenotyping data. Attention to these steps is key in the analysis of lymphoma samples,

ro

which pose challenges related to high cell density and lack of clear boundaries between tumor and stroma. Here we describe a detailed workflow for automated staining and quantification of

-p

immune cell subsets in DLBCL.

re

Automated image analysis, and a combination of machine learning and conventional image analysis algorithms, are all essential to ensure consistency while processing large number of

lP

images. Importantly, we highlight the use of a binary phenotyping approach, which offers some key advantages over commonly used systems that detect multiple markers simultaneously. Multi-

na

marker detection often requires the image analyst to create certain rules that may bias the results. For instance, some image analysis workflows require the user to first classify immune cells into

ur

either CD20 or CD3 buckets, only then to analyze subsets of these larger populations. Methods

Jo

that use bucketing of this type do not allow detection of dual CD20/CD3 cells, which are rare in the normal lymphoid system but may occur in lymphoma (Schuh et al., 2016; Yokose et al., 2001). Using binary phenotyping we are able to identify immune cells expressing multiple markers by overlapping individual marker phenomaps. The simplicity of the binary classification is also advantageous during algorithm training and verification of results. However, the binary approach demands more computational resources because it requires separate image analysis steps for each marker, and a second tier analysis for marker overlap in Matlab. The implementation of parallel processing can improve analysis speed considerably and is being evaluated in our lab. In summary, the workflow described here allows for automated, high throughput multiplex IF staining, and accurate immune phenotyping of lymphoma samples. The quantification of cells in the immune microenvironment of DLBCL and further understanding of

Journal Pre-proof their spatial features may shed light in the biology of this disease in the context of

Jo

ur

na

lP

re

-p

ro

of

immunotherapies.

Journal Pre-proof

Jo

ur

na

lP

re

-p

ro

of

Declarations of Interest: The authors declare no conflicts of interest.

Journal Pre-proof References Binnewies, M., Roberts, E. W., Kersten, K., Chan, V., Fearon, D. F., Merad, M., Coussens, L. M., Gabrilovich, D. I., Ostrand-Rosenberg, S., Hedrick, C. C., et al. (2018). Understanding the tumor immune microenvironment (TIME) for effective therapy. Nat Med 24, 541-550.

of

C-W Lee, E. W. S., S.D. Li, B. Liao (2010). Using gene-pair differential expression to assess relative contributions of biological processes in defining disease phenotype. In A Practical Guide to Bioinformatics Analysis, G.P.C. Fung, ed. (CreateSpace Independent Publishing Platform). Chao, J., DeBiasio, R., Zhu, Z., Giuliano, K. A., and Schmidt, B. F. (1996). Immunofluorescence signal amplification by the enzyme-catalyzed deposition of a fluorescent reporter substrate (CARD). Cytometry 23, 48-53.

ro

Dixon, A. R., Bathany, C., Tsuei, M., White, J., Barald, K. F., and Takayama, S. (2015). Recent developments in multiplexing techniques for immunohistochemistry. Expert Rev Mol Diagn 15, 11711186.

re

-p

Dunphy, C. H. (2004). Applications of flow cytometry and immunohistochemistry to diagnostic hematopathology. Arch Pathol Lab Med 128, 1004-1022.

lP

Finak, G., Langweiler, M., Jaimes, M., Malek, M., Taghiyar, J., Korin, Y., Raddassi, K., Devine, L., Obermoser, G., Pekalski, M. L., et al. (2016). Standardizing Flow Cytometry Immunophenotyping Analysis from the Human ImmunoPhenotyping Consortium. Sci Rep 6, 20686.

na

Fiore, C., Bailey, D., Conlon, N., Wu, X., Martin, N., Fiorentino, M., Finn, S., Fall, K., Andersson, S. O., Andren, O., et al. (2012). Utility of multispectral imaging in automated quantitative scoring of immunohistochemistry. J Clin Pathol 65, 496-502.

ur

Friedman, J., Hastie, T., and Tibshirani, R. (2010). Regularization Paths for Generalized Linear Models via Coordinate Descent. J Stat Softw 33, 1-22.

Jo

Gorris, M. A. J., Halilovic, A., Rabold, K., van Duffelen, A., Wickramasinghe, I. N., Verweij, D., Wortel, I. M. N., Textor, J. C., de Vries, I. J. M., and Figdor, C. G. (2018). Eight-Color Multiplex Immunohistochemistry for Simultaneous Detection of Multiple Immune Checkpoint Molecules within the Tumor Microenvironment. J Immunol 200, 347-354. Hamid, O., Robert, C., Daud, A., Hodi, F. S., Hwu, W. J., Kefford, R., Wolchok, J. D., Hersey, P., Joseph, R. W., Weber, J. S., et al. (2013). Safety and tumor responses with lambrolizumab (anti-PD-1) in melanoma. N Engl J Med 369, 134-144. Kather, J. N., Suarez-Carmona, M., Charoentong, P., Weis, C. A., Hirsch, D., Bankhead, P., Horning, M., Ferber, D., Kel, I., Herpel, E., et al. (2018). Topography of cancer-associated immune cells in human solid tumors. Elife 7. Koelzer, V. H., Sirinukunwattana, K., Rittscher, J., and Mertz, K. D. (2019). Precision immunoprofiling by image analysis and artificial intelligence. Virchows Arch 474, 511-522.

Journal Pre-proof Metrock, L. K., Summers, R. J., Park, S., Gillespie, S., Castellino, S., Lew, G., and Keller, F. G. (2017). Utility of peripheral blood immunophenotyping by flow cytometry in the diagnosis of pediatric acute leukemia. Pediatr Blood Cancer 64. Morihiro, T., Kuroda, S., Kanaya, N., Kakiuchi, Y., Kubota, T., Aoyama, K., Tanaka, T., Kikuchi, S., Nagasaka, T., Nishizaki, M., et al. (2019). PD-L1 expression combined with microsatellite instability/CD8+ tumor infiltrating lymphocytes as a useful prognostic biomarker in gastric cancer. Sci Rep 9, 4633. Parra, E. R., Uraoka, N., Jiang, M., Cook, P., Gibbons, D., Forget, M. A., Bernatchez, C., Haymaker, C., Wistuba, II, and Rodriguez-Canales, J. (2017). Validation of multiplex immunofluorescence panels using multispectral microscopy for immune-profiling of formalin-fixed and paraffin-embedded human tumor tissues. Sci Rep 7, 13380.

-p

ro

of

Rosenwald, A., Wright, G., Chan, W. C., Connors, J. M., Campo, E., Fisher, R. I., Gascoyne, R. D., MullerHermelink, H. K., Smeland, E. B., Giltnane, J. M., et al. (2002). The use of molecular profiling to predict survival after chemotherapy for diffuse large-B-cell lymphoma. N Engl J Med 346, 1937-1947. Ryan Rifkin, A. K. (2004). In Defense of One-Vs-All Classification. Journal of Machine Learning Research 5, 101-141.

lP

re

Schuh, E., Berer, K., Mulazzani, M., Feil, K., Meinl, I., Lahm, H., Krane, M., Lange, R., Pfannes, K., Subklewe, M., et al. (2016). Features of Human CD3+CD20+ T Cells. J Immunol 197, 1111-1117. Schwen, L. O., Andersson, E., Korski, K., Weiss, N., Haase, S., Gaire, F., Hahn, H. K., Homeyer, A., and Grimm, O. (2018). Data-Driven Discovery of Immune Contexture Biomarkers. Front Oncol 8, 627. Stack, E. C., Wang, C., Roman, K. A., and Hoyt, C. C. (2014). Multiplexed immunohistochemistry, imaging, and quantitation: a review, with an assessment of Tyramide signal amplification, multispectral imaging and multiplex analysis. Methods 70, 46-58.

ur

na

Topalian, S. L., Hodi, F. S., Brahmer, J. R., Gettinger, S. N., Smith, D. C., McDermott, D. F., Powderly, J. D., Carvajal, R. D., Sosman, J. A., Atkins, M. B., et al. (2012). Safety, activity, and immune correlates of antiPD-1 antibody in cancer. N Engl J Med 366, 2443-2454.

Jo

Tsujikawa, T., Kumar, S., Borkar, R. N., Azimi, V., Thibault, G., Chang, Y. H., Balter, A., Kawashima, R., Choe, G., Sauer, D., et al. (2017). Quantitative Multiplex Immunohistochemistry Reveals MyeloidInflamed Tumor-Immune Complexity Associated with Poor Prognosis. Cell Rep 19, 203-217. van Gijlswijk, R. P., Zijlmans, H. J., Wiegant, J., Bobrow, M. N., Erickson, T. J., Adler, K. E., Tanke, H. J., and Raap, A. K. (1997). Fluorochrome-labeled tyramides: use in immunocytochemistry and fluorescence in situ hybridization. J Histochem Cytochem 45, 375-382. Yokose, N., Ogata, K., Sugisaki, Y., Mori, S., Yamada, T., An, E., and Dan, K. (2001). CD20-positive T cell leukemia/lymphoma: case report and review of the literature. Ann Hematol 80, 372-375.

Journal Pre-proof

na

lP

re

-p

ro

of

Supplementary montage

ur

Singleplex IHC evaluation of an antibody (anti-CD47) in positive and negative cell pellet controls made of Chinese hamster ovary (CHO) cells transfected (a) or not (b) with the target of

Jo

interest. (c), (d), and (e) are micrographs of 3 different staining conditions applied on serial sections of a representative tissue sample. (f) displays the ISH staining performed on a serial section. Scale bare = 50 m.

Journal Pre-proof Supplementary Protocol

BOND RX Step Reagent

Supplier: Not applicable

1 Buffer Inc. (min): 0:00

Temperature: Ambient

of

Step type: Reagent

ro

Step Reagent 2 *Bond Wash Solution Inc. (min): 0:00

Inc. (min): 0:00

Step Reagent

Step Reagent

Jo

5 *Open 1

Inc. (min): 5:00

Step type: Reagent

Inc. (min): 30:00

Dispense type: 150 µL

Supplier: Perkin Elmer

Temperature: Ambient

ur

Step type: Reagent

na

4 *PKI Blocking Buffer

Supplier: Leica Microsystems

Temperature: Ambient

lP

Step type: Wash

Dispense type: 150 µL

re

Step Reagent 3 *Bond Wash Solution

Supplier: Leica Microsystems

Temperature: Ambient

-p

Step type: Wash

Dispense type: 150 µL

Dispense type: 150 µL

Supplier: User

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

6 *Bond Wash Solution Step type: Wash

Inc. (min): 0:00

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

7 *Bond Wash Solution Step type: Wash

Inc. (min): 1:00

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

8 *Bond Wash Solution Step type: Wash

Inc. (min): 0:00

Temperature: Ambient

Dispense type: 150 µL

1/19 For research use only. Not for use in clinical procedures.

Journal Pre-proof C r e a t e d b y : C r e a t i o n t i m e : F a c i l i t y : S t a i n i n g s t a t u s :

Step Reagent

Supplier: Perkin Elmer

9 *Opal Polymer HRP Inc. (min): 10:00

Temperature: Ambient

Step Reagent

Supplier: Leica Microsystems

Temperature: Ambient

-p

Inc. (min): 0:00

ro

10 *Bond Wash Solution Step type: Wash

11 *Bond Wash Solution

Step Reagent

Step type: Wash

na

12 *Bond Wash Solution

Inc. (min): 0:00

ur

Step Reagent

Supplier: Leica Microsystems

Temperature: Ambient

lP

Inc. (min): 1:00

Dispense type: 150 µL

re

Step Reagent

Step type: Wash

Dispense type: 150 µL

of

Step type: Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

Temperature: Ambient

Dispense type: Open

Supplier: Leica Microsystems

13 *Bond Wash Solution

Inc. (min): 0:00

Jo

Step type: Wash

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

14 *Bond Wash Solution Step type: Wash

Inc. (min): 0:00

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Perkin Elmer

15 *Opal 520 Reagent Step type: Reagent

Inc. (min): 10:00

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

16 *Bond Wash Solution Step type: Wash

Inc. (min): 0:00

Temperature: Ambient

Dispense type: 150 µL

2/ 19 For research use only. Not for use in clinical procedures.

Journal Pre-proof C r e a t e d b y : C r e a t i o n t i m e : F a c i l i t y : S t a i n i n g s t a t u s :

Step Reagent

Supplier: Leica Microsystems

17 *Bond Wash Solution Inc. (min): 1:00

Temperature: Ambient

Step Reagent

Supplier: Leica Microsystems

Temperature: Ambient

-p

Inc. (min): 0:00

ro

18 *Bond Wash Solution Step type: Wash

19 *Bond Wash Solution

Step Reagent

Step Reagent

Inc. (min): 0:00

ur

Step type: Reagent

na

20 *Bond ER Solution 1

Supplier: Leica Microsystems

Temperature: Ambient

lP

Inc. (min): 0:00

Dispense type: 150 µL

re

Step Reagent

Step type: Wash

Dispense type: 150 µL

of

Step type: Wash

Dispense type: 150 µL

Supplier: Leica Microsystems

Temperature: Ambient

Dispense type: 150 µL

Supplier: Leica Microsystems

21 *Bond ER Solution 1

Inc. (min): 0:00

Jo

Step type: Reagent

Temperature: 95

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

22 *Bond ER Solution 1 Step type: Reagent

Inc. (min): 20:00

Temperature: 95

Step Reagent

Dispense type: Intermediate

Supplier: Leica Microsystems

23 *Bond ER Solution 1 Step type: Reagent

Inc. (min): 0:00

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

24 *Bond Wash Solution Step type: Wash

Inc. (min): 0:00

Temperature: Ambient

Dispense type: 150 µL

3/ 19 For research use only. Not for use in clinical procedures.

Journal Pre-proof C r e a t e d b y : C r e a t i o n t i m e : F a c i l i t y : S t a i n i n g s t a t u s :

Step Reagent

Supplier: Leica Microsystems

25 *Bond Wash Solution Inc. (min): 1:00

Temperature: Ambient

Step Reagent

Supplier: Leica Microsystems

Temperature: Ambient

-p

Inc. (min): 0:00

ro

26 *Bond Wash Solution Step type: Wash

27 *PKI Blocking Buffer

Step Reagent

Step type: Reagent

Inc. (min): 30:00

ur

Step Reagent

na

28 *Open 2

Supplier: Perkin Elmer

Temperature: Ambient

lP

Inc. (min): 5:00

Dispense type: 150 µL

re

Step Reagent

Step type: Reagent

Dispense type: 150 µL

of

Step type: Wash

Dispense type: 150 µL

Supplier: User

Temperature: Ambient

Dispense type: 150 µL

Supplier: Leica Microsystems

29 *Bond Wash Solution

Inc. (min): 0:00

Jo

Step type: Wash

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

30 *Bond Wash Solution Step type: Wash

Inc. (min): 1:00

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

31 *Bond Wash Solution Step type: Wash

Inc. (min): 0:00

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Perkin Elmer

32 *Opal Polymer HRP Step type: Reagent

Inc. (min): 10:00

Temperature: Ambient

Dispense type: 150 µL

4/ 19 For research use only. Not for use in clinical procedures.

Journal Pre-proof C r e a t e d b y : C r e a t i o n t i m e : F a c i l i t y : S t a i n i n g s t a t u s :

Step Reagent

Supplier: Leica Microsystems

33 *Bond Wash Solution Inc. (min): 0:00

Temperature: Ambient

Step Reagent

Supplier: Leica Microsystems

Temperature: Ambient

-p

Inc. (min): 1:00

ro

34 *Bond Wash Solution Step type: Wash

35 *Bond Wash Solution

Step Reagent

Step type: Wash

na

36 *Bond Wash Solution

Inc. (min): 0:00

ur

Step Reagent

Supplier: Leica Microsystems

Temperature: Ambient

lP

Inc. (min): 0:00

Dispense type: 150 µL

re

Step Reagent

Step type: Wash

Dispense type: 150 µL

of

Step type: Wash

Dispense type: Open

Supplier: Leica Microsystems

Temperature: Ambient

Dispense type: 150 µL

Supplier: Leica Microsystems

37 *Bond Wash Solution

Inc. (min): 0:00

Jo

Step type: Wash

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Perkin Elmer

38 *Opal 540 Reagent Step type: Reagent

Inc. (min): 10:00

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

39 *Bond Wash Solution Step type: Wash

Inc. (min): 0:00

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

40 *Bond Wash Solution Step type: Wash

Inc. (min): 1:00

Temperature: Ambient

For research use only. Not for use in clinical procedures.

Dispense type: 150 µL

Journal Pre-proof C r e a t e d b y : C r e a t i o n t i m e : F a c i l i t y : S t a i n i n g s t a t u s :

Step Reagent

Supplier: Leica Microsystems

41 *Bond Wash Solution Inc. (min): 0:00

Temperature: Ambient

Step Reagent

Supplier: Leica Microsystems

Temperature: Ambient

-p

Inc. (min): 0:00

ro

42 *Bond Wash Solution Step type: Wash

43 *Bond ER Solution 1

Step Reagent

Step Reagent

Inc. (min): 0:00

ur

Step type: Reagent

na

44 *Bond ER Solution 1

Supplier: Leica Microsystems

Temperature: Ambient

lP

Inc. (min): 0:00

Dispense type: 150 µL

re

Step Reagent

Step type: Reagent

Dispense type: 150 µL

of

Step type: Wash

Dispense type: 150 µL

Supplier: Leica Microsystems

Temperature: 95

Dispense type: 150 µL

Supplier: Leica Microsystems

45 *Bond ER Solution 1

Inc. (min): 20:00

Jo

Step type: Reagent

Temperature: 95

Step Reagent

Dispense type: Intermediate

Supplier: Leica Microsystems

46 *Bond ER Solution 1 Step type: Reagent

Inc. (min): 0:00

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

47 *Bond Wash Solution Step type: Wash

Inc. (min): 0:00

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

48 *Bond Wash Solution Step type: Wash

Inc. (min): 1:00

Temperature: Ambient

Dispense type: 150 µL

6/ 19 For research use only. Not for use in clinical procedures.

Journal Pre-proof C r e a t e d b y : C r e a t i o n t i m e : F a c i l i t y : S t a i n i n g s t a t u s :

Step Reagent

Supplier: Leica Microsystems

49 *Bond Wash Solution Inc. (min): 0:00

Temperature: Ambient

Step Reagent

Supplier: Perkin Elmer

Temperature: Ambient

-p

Inc. (min): 5:00

ro

50 *PKI Blocking Buffer Step type: Reagent

51 *Open 3

Step Reagent

Step type: Wash

na

52 *Bond Wash Solution

Inc. (min): 0:00

ur

Step Reagent

Supplier: User

Temperature: Ambient

lP

Inc. (min): 30:00

Dispense type: 150 µL

re

Step Reagent

Step type: Reagent

Dispense type: 150 µL

of

Step type: Wash

Dispense type: 150 µL

Supplier: Leica Microsystems

Temperature: Ambient

Dispense type: 150 µL

Supplier: Leica Microsystems

53 *Bond Wash Solution

Inc. (min): 1:00

Jo

Step type: Wash

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

54 *Bond Wash Solution Step type: Wash

Inc. (min): 0:00

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Perkin Elmer

55 *Opal Polymer HRP Step type: Reagent

Inc. (min): 10:00

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

56 *Bond Wash Solution Step type: Wash

Inc. (min): 0:00

Temperature: Ambient

Dispense type: 150 µL

7/ 19 For research use only. Not for use in clinical procedures.

Journal Pre-proof C r e a t e d b y : C r e a t i o n t i m e : F a c i l i t y : S t a i n i n g s t a t u s :

Step Reagent

Supplier: Leica Microsystems

57 *Bond Wash Solution Inc. (min): 1:00

Temperature: Ambient

Step Reagent

Supplier: Leica Microsystems

Temperature: Ambient

-p

Inc. (min): 0:00

ro

58 *Bond Wash Solution Step type: Wash

59 *Bond Wash Solution

Step Reagent

Inc. (min): 0:00

Step Reagent

ur

Step type: Wash

na

60 *Bond Wash Solution

Supplier: Leica Microsystems

Temperature: Ambient

lP

Inc. (min): 0:00

Dispense type: Open

re

Step Reagent

Step type: Wash

Dispense type: 150 µL

of

Step type: Wash

Dispense type: 150 µL

Supplier: Leica Microsystems

Temperature: Ambient

Dispense type: 150 µL

Supplier: Perkin Elmer

61 *Opal 690 Reagent

Inc. (min): 10:00

Jo

Step type: Reagent

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

62 *Bond Wash Solution Step type: Wash

Inc. (min): 0:00

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

63 *Bond Wash Solution Step type: Wash

Inc. (min): 1:00

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

64 *Bond Wash Solution Step type: Wash

Inc. (min): 0:00

Temperature: Ambient

Dispense type: 150 µL

8/19 For research use only. Not for use in clinical procedures.

Journal Pre-proof C r e a t e d b y : C r e a t i o n t i m e : F a c i l i t y : S t a i n i n g s t a t u s :

Step Reagent

Supplier: Leica Microsystems

65 *Bond Wash Solution Inc. (min): 0:00

Temperature: Ambient

Step Reagent

Supplier: Leica Microsystems

Temperature: Ambient

-p

Inc. (min): 0:00

ro

66 *Bond ER Solution 1 Step type: Reagent

67 *Bond ER Solution 1

Step Reagent

Step Reagent

Inc. (min): 20:00

ur

Step type: Reagent

na

68 *Bond ER Solution 1

Supplier: Leica Microsystems

Temperature: 95

lP

Inc. (min): 0:00

Dispense type: 150 µL

re

Step Reagent

Step type: Reagent

Dispense type: 150 µL

of

Step type: Wash

Dispense type: 150 µL

Supplier: Leica Microsystems

Temperature: 95

Dispense type: Intermediate

Supplier: Leica Microsystems

69 *Bond ER Solution 1

Inc. (min): 0:00

Jo

Step type: Reagent

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

70 *Bond Wash Solution Step type: Wash

Inc. (min): 0:00

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

71 *Bond Wash Solution Step type: Wash

Inc. (min): 1:00

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

72 *Bond Wash Solution Step type: Wash

Inc. (min): 0:00

Temperature: Ambient

Dispense type: 150 µL

9/ 19 For research use only. Not for use in clinical procedures.

Journal Pre-proof C r e a t e d b y : C r e a t i o n t i m e : F a c i l i t y : S t a i n i n g s t a t u s :

Step Reagent

Supplier: Perkin Elmer

73 *PKI Blocking Buffer Inc. (min): 5:00

Temperature: Ambient

Step Reagent

Supplier: User

Temperature: Ambient

-p

Inc. (min): 30:00

ro

74 *Open 4 Step type: Reagent

75 *Bond Wash Solution

Step Reagent

Step type: Wash

na

76 *Bond Wash Solution

Inc. (min): 1:00

ur

Step Reagent

Supplier: Leica Microsystems

Temperature: Ambient

lP

Inc. (min): 0:00

Dispense type: 150 µL

re

Step Reagent

Step type: Wash

Dispense type: 150 µL

of

Step type: Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

Temperature: Ambient

Dispense type: 150 µL

Supplier: Leica Microsystems

77 *Bond Wash Solution

Inc. (min): 0:00

Jo

Step type: Wash

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Perkin Elmer

78 *Opal Polymer HRP Step type: Reagent

Inc. (min): 10:00

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

79 *Bond Wash Solution Step type: Wash

Inc. (min): 0:00

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

80 *Bond Wash Solution Step type: Wash

Inc. (min): 1:00

Temperature: Ambient

Dispense type: 150 µL

10/19 For research use only. Not for use in clinical procedures.

Journal Pre-proof C r e a t e d b y : C r e a t i o n t i m e : F a c i l i t y : S t a i n i n g s t a t u s :

Step Reagent

Supplier: Leica Microsystems

81 *Bond Wash Solution Inc. (min): 0:00

Temperature: Ambient

Step Reagent

Supplier: Leica Microsystems

Temperature: Ambient

-p

Inc. (min): 0:00

ro

82 *Bond Wash Solution Step type: Wash

83 *Bond Wash Solution

Step Reagent

Step type: Reagent

Inc. (min): 10:00

ur

Step Reagent

na

84 *Opal 570 Reagent

Supplier: Leica Microsystems

Temperature: Ambient

lP

Inc. (min): 0:00

Dispense type: 150 µL

re

Step Reagent

Step type: Wash

Dispense type: Open

of

Step type: Wash

Dispense type: 150 µL

Supplier: Perkin Elmer

Temperature: Ambient

Dispense type: 150 µL

Supplier: Leica Microsystems

85 *Bond Wash Solution

Inc. (min): 0:00

Jo

Step type: Wash

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

86 *Bond Wash Solution Step type: Wash

Inc. (min): 1:00

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

87 *Bond Wash Solution Step type: Wash

Inc. (min): 0:00

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

88 *Bond Wash Solution Step type: Wash

Inc. (min): 0:00

Temperature: Ambient

Dispense type: 150 µL

11/19 For research use only. Not for use in clinical procedures.

Journal Pre-proof C r e a t e d b y : C r e a t i o n t i m e : F a c i l i t y : S t a i n i n g s t a t u s :

Step Reagent

Supplier: Leica Microsystems

89 *Bond ER Solution 1 Inc. (min): 0:00

Temperature: Ambient

Step Reagent

Supplier: Leica Microsystems

Temperature: 95

-p

Inc. (min): 0:00

ro

90 *Bond ER Solution 1 Step type: Reagent

91 *Bond ER Solution 1

Step Reagent

Step type: Reagent

Inc. (min): 0:00

ur

Step Reagent

na

92 *Bond ER Solution 1

Supplier: Leica Microsystems

Temperature: 95

lP

Inc. (min): 20:00

Dispense type: 150 µL

re

Step Reagent

Step type: Reagent

Dispense type: 150 µL

of

Step type: Reagent

Dispense type: Intermediate

Supplier: Leica Microsystems

Temperature: Ambient

Dispense type: 150 µL

Supplier: Leica Microsystems

93 *Bond Wash Solution

Inc. (min): 0:00

Jo

Step type: Wash

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

94 *Bond Wash Solution Step type: Wash

Inc. (min): 1:00

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

95 *Bond Wash Solution Step type: Wash

Inc. (min): 0:00

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Perkin Elmer

96 *PKI Blocking Buffer Step type: Reagent

Inc. (min): 5:00

Temperature: Ambient

Dispense type: 150 µL

12/19 For research use only. Not for use in clinical procedures.

Journal Pre-proof C r e a t e d b y : C r e a t i o n t i m e : F a c i l i t y : S t a i n i n g s t a t u s :

Step Reagent

Supplier: User

97 *Open 5 Inc. (min): 30:00

Temperature: Ambient

Step Reagent

Supplier: Leica Microsystems

Temperature: Ambient

-p

Inc. (min): 0:00

ro

98 *Bond Wash Solution Step type: Wash

99 *Bond Wash Solution

Step Reagent

Inc. (min): 0:00

Step Reagent

ur

Step type: Wash

na

100 *Bond Wash Solution

Supplier: Leica Microsystems

Temperature: Ambient

lP

Inc. (min): 1:00

Dispense type: 150 µL

re

Step Reagent

Step type: Wash

Dispense type: 150 µL

of

Step type: Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

Temperature: Ambient

Dispense type: 150 µL

Supplier: Perkin Elmer

101 *Opal Polymer HRP

Inc. (min): 10:00

Jo

Step type: Reagent

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

102 *Bond Wash Solution Step type: Wash

Inc. (min): 0:00

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

103 *Bond Wash Solution Step type: Wash

Inc. (min): 1:00

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

104 *Bond Wash Solution Step type: Wash

Inc. (min): 0:00

Temperature: Ambient

Dispense type: Open

13/19 For research use only. Not for use in clinical procedures.

Journal Pre-proof C r e a t e d b y : C r e a t i o n t i m e : F a c i l i t y : S t a i n i n g s t a t u s :

Step Reagent

Supplier: Leica Microsystems

105 *Bond Wash Solution Inc. (min): 0:00

Temperature: Ambient

Step Reagent

Supplier: Leica Microsystems

Temperature: Ambient

-p

Inc. (min): 0:00

ro

106 *Bond Wash Solution Step type: Wash

107 *Opal 620 Reagent

Step Reagent

Step type: Wash

na

108 *Bond Wash Solution

Inc. (min): 0:00

ur

Step Reagent

Supplier: Perkin Elmer

Temperature: Ambient

lP

Inc. (min): 10:00

Dispense type: 150 µL

re

Step Reagent

Step type: Reagent

Dispense type: 150 µL

of

Step type: Wash

Dispense type: 150 µL

Supplier: Leica Microsystems

Temperature: Ambient

Dispense type: 150 µL

Supplier: Leica Microsystems

109 *Bond Wash Solution

Inc. (min): 1:00

Jo

Step type: Wash

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

110 *Bond Wash Solution Step type: Wash

Inc. (min): 0:00

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

111 *Bond Wash Solution Step type: Wash

Inc. (min): 0:00

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

112 *Bond ER Solution 1 Step type: Reagent

Inc. (min): 0:00

Temperature: Ambient

Dispense type: 150 µL

14/19 For research use only. Not for use in clinical procedures.

Journal Pre-proof C r e a t e d b y : C r e a t i o n t i m e : F a c i l i t y : S t a i n i n g s t a t u s :

Step Reagent

Supplier: Leica Microsystems

113 *Bond ER Solution 1 Inc. (min): 0:00

Temperature: 95

Step Reagent

Supplier: Leica Microsystems

Temperature: 95

-p

Inc. (min): 20:00

ro

114 *Bond ER Solution 1 Step type: Reagent

115 *Bond ER Solution 1

Step Reagent

Step type: Wash

na

116 *Bond Wash Solution

Inc. (min): 0:00

ur

Step Reagent

Supplier: Leica Microsystems

Temperature: Ambient

lP

Inc. (min): 0:00

Dispense type: Intermediate

re

Step Reagent

Step type: Reagent

Dispense type: 150 µL

of

Step type: Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

Temperature: Ambient

Dispense type: 150 µL

Supplier: Leica Microsystems

117 *Bond Wash Solution

Inc. (min): 1:00

Jo

Step type: Wash

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

118 *Bond Wash Solution Step type: Wash

Inc. (min): 0:00

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Perkin Elmer

119 *PKI Blocking Buffer Step type: Reagent

Inc. (min): 5:00

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: User

120 *Open 6 Step type: Reagent

Inc. (min): 30:00

Temperature: Ambient

Dispense type: 150 µL

15/19 For research use only. Not for use in clinical procedures.

Journal Pre-proof C r e a t e d b y : C r e a t i o n t i m e : F a c i l i t y : S t a i n i n g s t a t u s :

Step Reagent

Supplier: Leica Microsystems

121 *Bond Wash Solution Inc. (min): 0:00

Temperature: Ambient

Step Reagent

Supplier: Leica Microsystems

Temperature: Ambient

-p

Inc. (min): 1:00

ro

122 *Bond Wash Solution Step type: Wash

123 *Bond Wash Solution

Step Reagent

Step type: Reagent

Inc. (min): 10:00

ur

Step Reagent

na

124 *Opal Polymer HRP

Supplier: Leica Microsystems

Temperature: Ambient

lP

Inc. (min): 0:00

Dispense type: 150 µL

re

Step Reagent

Step type: Wash

Dispense type: 150 µL

of

Step type: Wash

Dispense type: 150 µL

Supplier: Perkin Elmer

Temperature: Ambient

Dispense type: 150 µL

Supplier: Leica Microsystems

125 *Bond Wash Solution

Inc. (min): 0:00

Jo

Step type: Wash

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

126 *Bond Wash Solution Step type: Wash

Inc. (min): 1:00

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

127 *Bond Wash Solution Step type: Wash

Inc. (min): 0:00

Temperature: Ambient

Step Reagent

Dispense type: Open

Supplier: Leica Microsystems

128 *Bond Wash Solution Step type: Wash

Inc. (min): 0:00

Temperature: Ambient

Dispense type: 150 µL

16/19 For research use only. Not for use in clinical procedures.

Journal Pre-proof C r e a t e d b y : C r e a t i o n t i m e : F a c i l i t y : S t a i n i n g s t a t u s :

Step Reagent

Supplier: Leica Microsystems

129 *Bond Wash Solution Inc. (min): 0:00

Temperature: Ambient

Step Reagent

Supplier: Perkin Elmer

Temperature: Ambient

-p

Inc. (min): 10:00

ro

130 *Opal 650 Reagent Step type: Reagent

131 *Bond Wash Solution

Step Reagent

Step type: Wash

na

132 *Bond Wash Solution

Inc. (min): 1:00

ur

Step Reagent

Supplier: Leica Microsystems

Temperature: Ambient

lP

Inc. (min): 0:00

Dispense type: 150 µL

re

Step Reagent

Step type: Wash

Dispense type: 150 µL

of

Step type: Wash

Dispense type: 150 µL

Supplier: Leica Microsystems

Temperature: Ambient

Dispense type: 150 µL

Supplier: Leica Microsystems

133 *Bond Wash Solution

Inc. (min): 0:00

Jo

Step type: Wash

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

134 *Bond Wash Solution Step type: Wash

Inc. (min): 0:00

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

135 *Bond ER Solution 1 Step type: Reagent

Inc. (min): 0:00

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

136 *Bond ER Solution 1 Step type: Reagent

Inc. (min): 0:00

Temperature: 95

Dispense type: 150 µL

17/19 For research use only. Not for use in clinical procedures.

Journal Pre-proof C r e a t e d b y : C r e a t i o n t i m e : F a c i l i t y : S t a i n i n g s t a t u s :

Step Reagent

Supplier: Leica Microsystems

137 *Bond ER Solution 1 Inc. (min): 20:00

Temperature: 95

Dispense type: Intermediate

of

Step type: Reagent

Step Reagent

Supplier: Leica Microsystems

Inc. (min): 0:00

Temperature: Ambient

-p

Step type: Reagent

ro

138 *Bond ER Solution 1

139 *Bond Wash Solution Inc. (min): 0:00

Step Reagent

na

140 *Bond Wash Solution Step type: Wash

Inc. (min): 1:00

ur

Step Reagent

Temperature: Ambient

lP

Step type: Wash

Supplier: Leica Microsystems

re

Step Reagent

Dispense type: 150 µL

Dispense type: 150 µL

Supplier: Leica Microsystems

Temperature: Ambient

Dispense type: 150 µL

Supplier: Leica Microsystems

141 *Bond Wash Solution

Inc. (min): 0:00

Jo

Step type: Wash

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Perkin Elmer

142 *Spectral DAPI Step type: Reagent

Inc. (min): 5:00

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

143 *Bond Wash Solution Step type: Wash

Inc. (min): 0:00

Temperature: Ambient

Step Reagent

Dispense type: 150 µL

Supplier: Leica Microsystems

144 *Bond Wash Solution Step type: Wash

Inc. (min): 1:00

Temperature: Ambient

Dispense type: 150 µL

18/19 For research use only. Not for use in clinical procedures.

Journal Pre-proof C r e a t e d b y : C r e a t i o n t i m e : F a c i l i t y : S t a i n i n g s t a t u s :

Step Reagent

Supplier: Leica Microsystems

145 *Bond Wash Solution Temperature: Ambient

Dispense type: 150 µL

ur

na

lP

re

-p

ro

of

Inc. (min): 0:00

Jo

Step type: Wash

19/19 For research use only. Not for use in clinical procedures.

Jo

ur

na

lP

re

-p

ro

of

Journal Pre-proof

Journal Pre-proof Highlights

ur

na

lP

re

-p

ro

of

Reliable multiplex immunofluorescence workflow is developed for high-throughput analysis. Accurate and consistent immune cell profiling algorithm is developed for lymphoma. Importance of spatial context in tumor microenvironment is demonstrated.

Jo

  