Anatomic pathology has a long history in modern medicine, dating back to Giovanni Morgagni (1682-1771), who is considered its father, and to Rudolf Virchow (1821-1902), who is considered the father of microscopic pathology. From those early days, the use of microscopy in anatomic pathology has evolved slowly. In the last decade digital imaging and computer analyses have ushered in a new era in pathology, beginning first in research, and now emerging into clinical practice.
For the most part, the vast majority of microscopic pathology is still done by viewing or imaging one marker at a time in serial sections, using a hematoxylin counterstain and either eosin (H&E) or, for protein IHC measurements, 3,3′-diaminobenzidine (DAB) as a chromogen. This one-stain-at-a-time method works well for measuring protein expression if all that is needed is an overall expression pattern of that protein in a tumor. Where a single-stain methodology breaks down is when one needs to know the expression level of more than one protein in the same cell or where there is an interest in how multiple proteins interact in single cells in situ.
Why is this important? Recently tumors have begun to be considered almost as organs in their own right, and as such their complexity can only be understood by studying the relationships among the specialized cell types which make up the tumor
microenvironment. Endothelial cells, pericytes, cancer-associated fibroblasts, cancer stem cells, and immune inflammatory cells are all specialized sub-populations of cells that are of importance in this tumor heterogeneity. In particular, and of importance in cancer immunotherapy, inflammatory cells can operate in conflicting ways, with both tumor-supporting and tumor-killing subclasses; the balance between the conflicting inflammatory responses in tumors is likely to prove instrumental in prognosis and, quite possibly, in therapies.1
For many application areas involving the tumor microenvironment, epithelial-mesenchymal transition (EMT), and cancer immunology, it is necessary to analyze complex multimarker phenotypes of cells in situ in tumor sections and to be able to understand their distribution and locations, both inside and outside the tumor.2 In the field of cancer immunology in general and cancer immunotherapy in particular, multimarker phenotyping of these subsets of immune cells is
Cancer immunotherapy is rapidly changing the landscape of cancer treatment, and monitoring the immune status of a patient is a critical aspect of research in this field. While quantifying the numbers of specific subsets of immune cells in blood is routine using multimarker flow cytometry, monitoring these same subsets of immune cells in solid tumors remains unobtainable with standard methods. Methods that involve decomposing the tumor and releasing the immune cells for flow cytometric analyses are difficult, and even when working perfectly, with no biasing of the cell populations, these methods cannot reveal the spatial relationships and distributions of the immune cells within and around the tumor. Standard IHC methodologies can be used on tissue sections to visualize or quantitate one marker at a time, but standard methods cannot capture the complex, multimarker phenotypes needed to analyze the subsets of immune cells that are important in cancer immunology.4,5
To address this issue a novel multimarker staining, imaging, and quantitation methodology has been developed that allows complex phenotypes to be determined and immune cells to be enumerated in situ in FFPE sections. The success of this application hinges on the bringing together of four methodologies, which combine to provide the per-cell quantitative results necessary.
First, a multimarker staining methodology enables the multiplexing in a single tissue section of up to four chromogenic immunostains in brightfield or up to eight immunostains in fluorescence. This means that a large number of immunology-specific markers can be multiplexed in a single panel (e.g., CD4, CD8, CD20, CD25, CD68, CD69, FOXP3, and Ki67, or any other combination of interest).
Second, a multispectral imaging (MSI) approach is used to enable the quantitative separation, or unmixing, of each of these markers into its own channel/image. Advanced multispectral imaging systems remove cross-talk between chromogens or fluorophores and are able to remove interfering tissue autofluorescence, thus providing quantitative data from which the per-cell intensity values are derived.6,7
In the third step, advanced, user-trainable morphologic image analysis software utilizes the multimarker, multispectral data to calculate per-cell and per-cellular-compartment (i.e., nuclear, cytoplasmic, and membrane) intensity values which are used to phenotype each cell (e.g., this is a CD4+/CD25+/FOXP3+ T cell). The analysis software also utilizes sophisticated machine learning to allow the user to train it to recognize and find morphologic regions of the tissue (e.g., tumor, dermis, inflammation, etc.) which then allow the determination of which cells are, for instance, in the tumor and which are in the stromal regions.8
The final piece to the puzzle is to add automation, enabling the walkaway imaging and subsequent analysis of the large volume of slides typically found in this type of study.9
A simple but useful example of this kind of “tissue cytometry” imaging can be seen in Figures 1 and 2. This tonsil section has been stained for CD4 (FITC), CD8 (Cy3), CD20 (Cy5), cytokeratin (coumarin) and nuclei (DAPI). It is a multiplexed panel that enables the simultaneous imaging of T helper cells (CD4), cytotoxic T cells (CD8), B cells (CD20), and epithelial cells (cytokeratin) in a single section. A multispectral image was acquired of this sample, which was then unmixed. Figure 1a shows the color (RGB) image of the sample as it looks down the microscope. Figure 1b shows the unmixed composite image, with DAPI (blue), CD4 (green), CD8 (magenta), CD20 (red), and cytokeratin (cyan) shown in pseudocolor. Figures 1c-1h show the individual unmixed images. Figure 2 shows the sample split into two halves: the lower left shows the unmixed composite image from Figure 1b; the upper right shows the result of the per-cell phenotyping analysis and is a map of the distributions of the various immune cells.
Further analyses of this kind of data can be performed, looking at densities, distributions, and relationships among the cells in the image, or to generate flow cytometry-like scatter plots that can show the relationships between clusters of cells in the graph and where those cells are in the imagery. Although this study is performed on tonsil as an example, it is easily applicable to sections from any solid tumor or with practically any combination of markers up to four immunostains in brightfield or up to eight markers in fluorescence.
As the questions being asked of biopsy specimens are becoming increasingly complex, the methods of analysis are having to adapt. Immunohistochemistry (IHC) is ubiquitous; the need for quantitative accuracy in IHC in brightfield and fluorescence (IF) is increasing. It is clear that an understanding of the tumor microenvironment is critical for tumor biology and for the development of new therapeutics. As we’ve learned more about the complexities of cancer, we’ve recognized the need to explore the distributions and relationships among the diverse cell types that comprise solid tumors. This has necessitated being able to make complex cellular phenotyping assessments, often involving multiple antigens, imaged and quantitated simultaneously in the same cell. For this reason, the adoption of multiplexed staining methods, multispectral imaging methodologies, and sophisticated image analysis strategies that can take advantage of the multimarker data is spreading rapidly. In the field of cancer immunology the ability to image and analyze the location and distributions of complex subsets of immune cells is of particular importance. All lymphocytes have the same external morphologic appearance; their function is revealed only through determining their multimarker phenotype. Now it is possible to perform this type of phenotyping analysis in situ in FFPE tumor sections.
- Hanahan D, Weinberg RA. Hallmarks of cancer: the next generation. Cell. 2011;144:646-674.
- Nicholson T, Sehgal PD, Drew SA, Huang W, Ricke WA. Sex steroid receptor expression and localization in benign prostatic hyperplasia varies with tissue compartment. Differentiation. 2013; 85:140-149.
- Setiadi AF, Ray NC, Kohrt HE, et al. Quantitative, architectural analysis of immune cell subsets in tumor-draining lymph nodes from breast cancer patients and healthy lymph nodes. PLoS ONE. 2010;5(8). http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0012420. Accessed May 12, 2014.
- de Boer OJ, van der Loos CM, Teeling P, van der Wal AC, Teunissen MB. Immunohistochemical analysis of regulatory T cell markers FOXP3 and GITR on CD4+CD25+ T cells in normal skin and inflammatory dermatoses. J. Histochem Cytochem. 2007;55:891-898.
- Nielsen JS, Sahota RA, Milne K, et al. CD20+ tumor-infiltrating lymphocytes have an atypical CD27−memory phenotype and together with CD8+ T cells promote favorable prognosis in ovarian cancer. Clin Cancer Res. 2012;18:3281-3292.
- Levenson RM, Mansfield JR. Multispectral imaging in biology and medicine: slices of life. Cytometry A. 2006;69:748-758.
- Mansfield JR, Hoyt C, Levenson RM. Visualization of microscopy-based spectral imaging data from multi-label tissue sections. Curr Prot Mol Biol. 2008;84:14.19.1–14.19.15.
- van der Loos CM, de Boer OJ, Mackaaij C, Hoekstra TL, van Gulik TM, Verheij J. Accurate quantitation of Ki67-positive proliferating hepatocytes in rabbit liver by a multicolor immunohistochemical (IHC) approach analyzed with automated tissue and cell segmentation software. J. Histochem Cytochem. 2013;61:11-18.
- Ding Z, Wu C-J, Chu GC, et al. SMAD-4-dependent barrier constrains prostate cancer growth and metastatic progression. Nature. 2011;470:269.