SITC 2022

Background The characterization of the tumor microenvironment is essential for a deep understanding of tumor biology and the identification of personalized targeted therapies. By means of new spatial proteomic approaches, a detailed mapping of the different immune cells has been made possible [1]. Among them, multiplex immunofluorescence has emerged as a key technology, maximizing the amount of extractable information from a cancer tissue section [2]. While lineage markers can usually be easily detected via canonical indirect immunofluorescence, functional markers often show much lower levels of expression and would benefit from a signal amplification strategy [3]. Combining the possibility to amplify low-expressed markers with a system capable of performing multiplex staining could provide significantly improved tumor immunoprofiling with detailed information on immune cell activation and exhaustion.
Here, we aim to implement amplification technologies on the COMET™ system, capable of performing automated multiplex immunofluorescence, to improve the detection of low-expressed markers. 

Methods The COMET™ platform is a microfluidic-based system that enables fully automated sequential immunofluorescent (seqIF™) assays based on iterative series of fast tissue staining, imaging, and antibody elution cycles on the same tissue section. Two amplification technologies were automated: Fluorescent Signal Amplification via Cyclic Staining of Target Molecules (FRACTAL) [4], and enzymatic Tyramide-based Signal Amplification (TSA) [5], also testing different commercially available TSA kits. The effects of amplification on low-expressed markers were evaluated with respect to standard assays on multiple human FFPE tissues. Normalized Mean Intensity and Signal to Background ratio have been used as parameters for the evaluation.   

Results The two amplification technologies, FRACTAL and TSA, were successfully automated and optimized on COMET™. The signal intensity of multiple challenging markers, including FoxP3, PD1, PD-L1, and IDO-1 – though being detectable by standard seqIF™ – was significantly amplified by the two techniques. Overall, FRACTAL appeared to be the most suitable technology as shown by the better performance in terms of signal amplification (by NMI or SBR quantification) over the TSA amplification tested on COMET™. Importantly, optimization of the elution step resulted in efficient removal of all antibody layers employed in the FRACTAL method from the stained tissues, therefore allowing its use in sequential staining cycles on the same section.  

Conclusions The implementation of amplification technologies in the COMET™ workflow for hyper-plex panel development will allow the detection of difficult-to-target biomarkers, such as functional markers with low-expression levels. This strategy can provide a significant improvement in immune cell profiling of tumor samples.  

Background: Spatial biology enables the interrogation of tissue composition at a single cell level with preservation of spatial context, which opens new avenues for tumor microenvironment (TME) studies [1]. Biomarkers’ composition of the tissue can be interrogated with hyperplexed immunofluorescence, wherein an imaging detection is performed for each marker on the same slide. The COMET™ platform performs sequential immunofluorescence (seqIF ™) and enables full automation of such workflow, where: up to 40 biomarkers can be detected with full automation from staining to data acquisition. The resulting hyperplex images are rich sources of data about the specimens. To extract information from such a dataset, Oncotopix® Discovery (Visiopharm) has a dedicated pipeline for image analysis that delivers single cell phenotypic information and biodistribution, providing access to the spatial composition of the TME. 

Methods: An FFPE Lung Tissue Microarray underwent sequential cycles of staining and imaging with COMET™ platform seqIF™ assay. Iterative cycles of staining, imaging and antibody elution allowed the detection of 20 antigens spanning across epithelial tumor and immune markers (panCK, E-Cad, aSMA, CD31, CD3, CD4, CD8, FoxP3, CD20, HLA-DR, Ki67, Vimentin, CD16, CD68, CD11b, CD163, CD14, CD11c, PD-1, PD-L1). The resulting image contains 23 channels: nuclear detection with DAPI, 2 channels of tissue autofluorescence (AF), and 20 marker channels. All layers were aligned and stitched into a single ome.tiff automatically by the COMET™ control software. Subsequent AF subtraction was performed in the Viewer by Lunaphore. The AF-subtracted image was analyzed using Oncotopix Discovery. The analysis pipeline consisted of a deep-learning (DL)-based tissue segmentation (tumor, stroma, necrosis, etc.), a pre-trained DL DAPI nuclear segmentation step, cellular phenotyping, and spatial metrics among the various cell types. 

Results: We interrogated tumor composition with the use of COMET™ platform and Oncotopix Discovery Software from Visiopharm.  Specific cellular phenotypes of interest are proliferating tumor cells, proliferating T cells, immunosuppressive macrophages and antigen-presenting cells and their interactions with a focus on the PD-(L)1 pathway.  

Conclusion:  The combination of hyperplex staining and advanced image analysis and in situ cellular phenotyping allows the identification of tissue composition. It is a crucial step for understanding and harnessing tissue biology. Being able to analyze the spatial distribution of specifically phenotyped cells in the TME enables to identify reliable biomarkers as predictive factors of response to therapies. 

References: 

1. Mund A, Brunner AD, Mann Unbiased spatial proteomics with single-cell resolution in tissues. Molecular Cell 2022; 82:2335-2349  

Background: Colorectal carcinoma (CRC) represents a major worldwide healthburden and shows an increasing incidence particularly in youngerpatients. The majority of metastatic CRC (microsatellite stable,MSS) do not respond to current immunotherapies in contrast to thesmall subset of microsatellite unstable (MSI) patients. However,responses to immunotherapy were recently observed in subsets ofprimary MSS tumors indicating their potential vulnerability to suchtherapeutic approaches. Here, we use multiplex imaging coupledwith powerful image analysis tools to highlight importantphenotypic differences between MSI and MSS.

Methods: 100 human CRC sections (30MSI, 70 MSS) were examined on the COMET™ platform (Lunaphore) with a multiplex sequentialimmunofluorescence (seqIF) panel of 12 biomarkers (CD3, CD31,CD4, CD8, FoxP3, TCF-1, TOX, EOMES, CK, CD45RO, S100,D2-40).

Results: Images were preprocessed by background subtraction and localcontrast enhancement [1]. Cell segmentation was done with theStardist algorithm [2]. Cell phenotypes (CT) were defined bythreshold-based classifiers. The epithelium was detected based onCK expression. The density of CT was assessed per area of interest.Cell neighborhoods (CN) were defined as the composition of the 10closest CTs within 300um. CNs were clustered into CN-classesusing DB-SCAN [3].
Cell segmentation and phenotyping algorithms resulted in precisecell detection (Fig. 1). We compared CT infiltration patterns in MSIvs MSS in the tumor-epithelium and epithelium-stroma interfaces.While we observed no significant differences in CT densities at theinterface, CD3 cell density was significantly higher in MSI vs MSSpatients in the tumor-epithelium (Fig. 2), with MSS moreheterogeneously distributed with CD3-high and CD3-low outliers.

Conclusions: Differences in cell interactions CRC patients were examined usingCN analysis. We observed a CN composed of CD3, CD8 cytotoxicT cells, CD3+CD8+ and tumor cells to be differentially enrichedbetween MSI and MSS patients (Fig. 3). A more precise characterization of the immune response to cancersis essential to leverage the advantages of immune modulations inthe treatment of cancers. We confirm here important differencesbetween MSI and MSS CRCs in part systematic due to the higherantigen load in MSI CRCs but also highlight heterogeneity amongstthe MSS tumors. Characterization of the phenotype of the T-lymphocyte population and its localization with respect to elementssuch as epithelial cells and tertiary lymphoid structures may help todefine both the prognosis of tumors and the possibility of a responseto immune checkpoint therapy.

References: [1] S. M. Pizer et al., Computer Vision, Graphics, and ImageProcessing 39, 1987.[2] U. Schmidt et al., MICCAI, 2018.[3] Martin Ester, Hans-Peter Kriegel, Jörg Sander, and Xiaowei Xu.1996. A density-based algorithm for discovering clusters in largespatial databases with noise. In Proceedings of the SecondInternational Conference on Knowledge Discovery and DataMining (KDD’96). AAAI Press, 226–231.[4] Massey, F. J. 1951. The Kolmogorov-Smirnov Test forGoodness of Fit. Journal of the American Statistical Association,46(253), 68–78. [5] Cohen, J. 1988. Statistical power analysis for the behavioralsciences (2nd ed.). Hillside, NJ: Lawrence Erlbaum Associates.

Background: Current colon cancer classification, prognostication, and therapy decisions are based mainly on cancer staging [1,2]. However, additional biomarkers are needed to improve patient stratification and complement treatment decision-making strategies. Tumor budding is recognized as an independent prognostic factor in a variety of solid cancers [3,4]. Tumor buds (TBs) are isolated single tumor cells or groups of up to four tumor cells, located both peri- and intra-tumorally. A higher tumor bud count correlates with poor prognosis in colorectal cancer (CRC), and it is hypothesized that a subset of TBs represents, at least in part, an Epithelial-Mesenchymal Transition (EMT) state [5]. To explore this hypothesis further, we developed a sequential immunofluorescence (seqIF™)-based panel for a more spatial approach to characterizing the TB and its microenvironment.  

Methods: Human CRC sections from different cohorts underwent initial pre-processing on PT Module™ (Thermo Fisher) followed by automated cycles of seqIF and imaging performed on COMET™ (Lunaphore Technologies). Hyperplex panels consisting of >20 protein biomarkers were generated using off-the-shelf antibodies and served to characterize the tumor-stroma interactions in a preliminary cohort of samples. In a second stage, the spatial screening was performed on larger cohorts of neoadjuvant treated vs. untreated CRC patient samples. The final characterization of TBs and their interaction with the surrounding milieu was performed by a downstream image analysis of the stained tissues. 

Results: With COMET™ we built an optimized panel including >20 biomarkers for characterization of the tumor and the surrounding stroma. The panel robustness was tested in 15 different CRC cases and resulted in the outstanding quality of the stainings with a high signal-to-background ratio and elution efficiency (> 98%) for all biomarkers. The developed panel allowed the visualization of inter- and intra-tumoral heterogeneity of stained cases and the identification of cell populations including immune cells, cancer-associated fibroblasts, and tumor cells. In line with the previously described EMT phenotype of TBs [6,7], early observations showed loss of EpCAM and E-Cadherin in a group of TBs verifying decreased epithelial phenotype (Figure1).

Conclusion: The >20-plex panels generated on COMET allowed (i) to discriminate TBs signatures within the depth of tumor-stroma interactions and (ii) to extract valuable TBs features in terms of marker expression. Next, crucial info about TBs phenotypes and their cellular neighborhood will be obtained through an unsupervised analysis approach. This will finally serve to better identify novel immunograms of these cellular entities, thus better defining their therapeutic potential for a personalized medicine approach.  

References: 

[1] Brierley, J. D., Gospodarowicz, M. K. & Wittekind, C. TNM Classification of Malignant Tumours. 8th edition. Wiley Blackwell, 2017. 

[2] Amin, M. B. et al. (eds). AJCC Cancer Staging Manual. 8th edition. Springer, 2017. 

[3] Berg, K.B and Schaeffer, D.F. Tumor budding as a standardized parameter in gastrointestinal carcinomas: more than just the colon. Mod. Pathol, 2018, 31, 862-872. 

[4] Almangush, A., Salo, T., Hagstrom, J. & Leivo, I. Tumour budding in head and neck squamous cell carcinoma – a systematic review. Histopathology, 2014, 65, 587-594.  

[5] Rogers, A. C. et al. Systematic review and meta-analysis of the impact of tumour budding in colorectal cancer. Br. J. Cancer, 2016, 115, 831–840. 

[6] Grigore, A. D. et al. Tumor budding: the name is EMT. Partial EMT. J. Clin. Med., 2016, 5(5), 51. 

[7] De Smedt, L. et al. Expression profiling of budding cells in colorectal cancer reveals an EMT-like phenotype and molecular subtype switching. Br. J. Cancer, 2017, 116, 58–65. 

Background: The tumor and its microenvironment are distinguished by highly heterogeneous cell types in dynamic evolution [1]. In the past decades, the adoption of single-cell RNA sequencing technologies improved our understanding of intra- and inter-patient variations, refining cancer diagnosis and targeted therapy, while still lacking spatial context information [2]. The spatial analysis of single-cell gene expression provides previously missing information on cell interactions, crucial for innovative treatment opportunities. However, the large-scale implementation of these assays is hampered by the lack of automated and user-friendly workflow solutions [3].
Here, we aim at automating an in situ transcriptomic assay and integrating it with a proteomic workflow on the COMET™ platform for easy and comprehensive mapping of the tumor microenvironment.  

Methods: COMET™ is a microfluidic-based instrument capable of automated sequential immunofluorescence (seqIF™) assays. The in situ RNA detection was based on DNA padlock probe hybridization and ligation, followed by rolling circle amplification and gene detection with fluorophore-tagged probes. A panel of probes targeting various immune-oncology genes was designed and the gene expression was evaluated in combination with a high-plex seqIF™ analysis.  

Results: The protocol for in situ RNA detection was automated and optimized on COMET™, resulting in specific detection of multiple transcripts (cell-type specific markers, transcription factors, secreted cytokines) on various tumor samples, with a signal quality that allows downstream computational analysis of gene expression signals. The transcriptomic assay was then integrated with a seqIF™ workflow for the co-detection of several target proteins for tumor microenvironment characterization on the same tissue section (Figure 1). Double RNA and protein analysis allowed to examine the expression of biomarkers for which no good antibody clones are available, validation of new antibodies by detecting RNA and protein co-localization, in addition to the identification of the cellular source of secreted molecules. Compared to the manual protocol, the combined automated workflow resulted in a drastic time reduction being based on iterative fast cycles of detection and imaging of two transcripts or proteins each, lasting approximately 30 and 40 minutes, respectively.  

Conclusions: We showed here that in situ spatial transcriptomics assays can be fully automated and combined with spatial proteomics on the COMET™ platform for a multi-omics approach with advantages in terms of time and complexity reduction with respect to manual protocols. This combinatorial automated detection of RNAs and proteins of pivotal biomarkers provides a powerful new tool for a simpler and better mapping of the tissue spatial context. 

References:

[1] Janiszewska M. The microcosmos of intratumor heterogeneity: the space-time of cancer evolution. Oncogene. 2020 Mar;39(10):2031-2039. 

[2] Rao A, Barkley D, França GS, Yanai I. Exploring tissue architecture using spatial transcriptomics. Nature. 2021 Aug;596(7871):211-220. 

[3] Asp M, Bergenstråhle J, Lundeberg J. Spatially Resolved Transcriptomes-Next Generation Tools for Tissue Exploration. Bioessays. 2020 Oct;42(10):e1900221.

Background: In health and disease, cells are interacting with their microenvironment, which shapes their homeostasis and response to any stimuli [1]. Deciphering the tissue architecture emerged as a crucial step to better understand tissue biology and harness it in a future therapeutic intervention [2].
Recently, advancements were made to interrogate the tissue spatial composition with the development of immunofluorescence-based hyperplex assays [3,4], allowing the investigation of dozens of biomarkers on a single tissue slide. There is a growing interest to apply hyperplex assay on fresh frozen sections (FS), yet harsh procedures used in manual protocols are detrimental to tissue morphology and limit the use of this application [5]. Furthermore, manual protocols are laborious and time-consuming, allowing to process a limited number of samples. 
Here, we describe the capability to perform automated hyperplex assays for up to 32 biomarkers on fragile FS tissues as a cogent case combining staining quality with tissue preservation.  

Methods: COMET™ platform allows performing sequential immunofluorescence (seqIF™) assays based on an iterative series of fast tissue staining, imaging, and antibody elution cycles. Fresh frozen tissue sections of human and murine origins were fixed and permeabilized prior to loading on the device for automated staining protocols. Multiplex assay was performed for up to 32 biomarkers on a single tissue slide. To assess the tissue morphology preservation, standard hematoxylin and eosin (H&E) staining was performed on FS freshly processed or post-staining protocol.  

Results: In this study, a sequential immunofluorescence protocol was automated and optimized on COMET™ for both human and mouse frozen tissue sections. Murine spleen, brain and lung tissues were stained with multiplex panels of up to 6 proteins and showed an accurate detection of both common immune and organ-specific biomarkers. For a deep characterization of human lung cancer, 32 biomarkers were simultaneously detected on a single FS with optimal staining and a total procedure time of 23 hours. An H&E staining performed on the slide retrieved from the platform after the multiplex protocol showed excellent preservation of the tissue architecture in comparison with an unprocessed FS slide.  

Conclusion: This work proves the feasibility of performing automated hyper-plex assays on a variety of delicate frozen samples with high-quality results.  With this new toolbox, we aim to support and hasten discovery studies across multiple research fields overcoming current limitations in tissue spatial biology.

References:

Background: In recent years, advanced melanoma treatment has improved dramatically thanks to the advent of immunotherapy. Particularly immune checkpoint inhibitors (ICI), in some patients, have demonstrated to improve long-term outcomes associated with limited toxicity. However, only a small population of patients achieve a durable response to therapy, owing to the lack of clinically validated predictive biomarkers (reviewed in [1]). The availability of improved predictive biomarkers may allow the identification of patients who will most benefit from ICI treatment and those who may be susceptible to immune-related adverse events. This difficulty in obtaining clinically relevant predictive biomarkers underscores the complexity of the immune system and the heterogeneity of the tumor microenvironment. In the present study, we take the first steps towards stratification of advanced melanoma patients who received a combination of ICI. We studied the predictive value of a multi-parameter spatial signature composed of lymphocyte-activation gene-3 (LAG-3), programmed death-ligand 1 (PD-L1) and cluster of differentiation 8 (CD8) in a retrospective cohort of patients with metastatic melanoma treated with ICI. 

Methods: In this retrospective study, from the biobank of the Istituto Nazionale Tumori IRCCS Fondazione G. Pascale, we recovered FFPE skin metastasis samples obtained from 10 melanoma patients with AJCC 8th edition stage IV [2] subsequently treated with combined ICI, enrolled from September 2016 to November 2017. According to the Response Evaluation Criteria in Solid Tumors (RECIST 1.1) [3], 4 patients achieved response to the treatment, while 6 patients were non-responders. A single FFPE tumor tissue of each patient was stained using LabSat® platform (Lunaphore Technologies) performing a Tyramide Signal Amplification multiplex immunohistochemistry staining for PD-L1, CD8, and LAG-3 expression, followed by DAPI counterstaining and whole slide fluorescent scanning. Single cell segmentation, phenotyping and quantification were performed in QuPath (0.3.0). The workflow is depicted in Figure 1. 

Results: Responder patients showed a statistically significant increase in CD8+ single-positive cell frequency compared to non-responders (Fig. 2A and 2C). Non-responder patients displayed a statistically significant increase of PD-L1+ single-positive cell frequency, as well as statistically significant increased frequency of double CD8+PD-L1+ positive cells, previously found to be a poor prognostic in multiple cancer types [4,5], and triple positive cells (Fig. 2B and 2C). 

Conclusion: We show here preliminary evidence of the predictive value of a spatial biomarker signature in patients who underwent combined ICI therapy for advanced melanoma. To further demonstrate clinical relevance, a more detailed analysis using larger retrospective and prospective cohorts is ongoing. 

References
[1] Garutti M, Bonin S, Buriolla S, Bertoli E, Pizzichetta MA, Zalaudek I, et al. Find the Flame: Predictive Biomarkers for Immunotherapy in Melanoma. Cancers 2021;13:1819. https://doi.org/10.3390/cancers13081819. 

[2] Gershenwald JE, Scolyer RA. Melanoma Staging: American Joint Committee on Cancer (AJCC) 8th Edition and Beyond. Ann. Surg. Oncol. 2018. p. 2105–10. 

[3] Eisenhauer EA, Therasse P, Bogaerts J, et al. New response evaluation criteria in solid tumors: revised RECIST guideline (version 1.1) Eur J Cancer. 2009;45:228–247. 

[4] Brochez L, Meireson A, Chevolet I, Sundahl N, Ost P, Kruse V. Challenging PD-L1 expressing cytotoxic T cells as a predictor for response to immunotherapy in melanoma. Nat Commun 2018;9:2921. https://doi.org/10.1038/s41467-018-05047-1. 

[5] Zhang L, Chen Y, Wang H, Xu Z, Wang Y, Li S, et al. Massive PD-L1 and CD8 double positive TILs characterize an immunosuppressive microenvironment with high mutational burden in lung cancer. J Immunother Cancer 2021;9:e002356. https://doi.org/10.1136/jitc-2021-002356.