A headache for cancer specialists
Janis Taube, a professor of pathology at Johns Hopkins University, focuses on microscopic interactions between immune cells and tumor cells as a way of predicting a patient's response to a particular treatment. She uses fluorescent stains to label specific cells or proteins so that she can observe cell-cell interactions. First, the signals from different fluorochromes can be superimposed on each other, which can affect the results. In addition, she usually looks at cell-cell interactions in the same plane, but the cells are actually in three dimensions, so she has to account for interactions in three dimensions as well.
Janis Taube, Professor of Pathology at Johns Hopkins UniversityIf you think that's not a big deal, you can see the complexity of the problem when you put it in the context of a tissue sample with millions of cells. Stitching together the interfering fluorescent signals into a complete image, and then determining how the cells are interacting in three dimensions, is driving Taube crazy. While she has no idea what to do with this problem, there is another group of experts at Johns Hopkins who deal with it on a daily basis.
Alexander Szalay is a professor in the departments of physics, astronomy, and computer science at Johns Hopkins, where he also chairs the Sloan Digital Sky SurveyNote 1 committee. His team works every day to stitch together millions of telescope images of billions of objects to create a 3D image of the universe. Do you see any similarities between this and Taube's research?
Alexander Szalay, Professor of Astronomy at Johns Hopkins UniversityUsing telescopes (microscopes) to look at many objects with different characteristics (different fluorescent signals), we can see that they have different characteristics. the position and interrelationships (cell-cell interactions) of many objects (millions of cells) in the sky (three-dimensional space).
Astronomy and tumor biology can actually learn from each other in the way they approach research and present information. Reference 1
On closer inspection, the imaging problems that Taube and Szalay have to deal with are actually similar. In some ways, though, astronomy is confronted with images that are more difficult to deal with than tumor biology. After all, objects vary with the seasons, and telescopes are subject to weather interference, but astronomers have managed to image the universe despite all these variables. In contrast, wouldn't it be easier to deal with slices of tissue that don't move and aren't disturbed by weather?
Thus, a collision of macro and micro, astronomy and tumor biology was born.1 Below, let's find out how this cross-disciplinary collaboration came about.
The utility of immunotherapy and the bottleneck of immunofluorescence stainingThe story begins with the PD-1 and PD-L1 blockers. blockers. Ever since scientists discovered that T-cells in the body attack tumor cells, they have been working on ways to activate them to destroy tumors, but tumor cells are not stupid.
There's a protein on the surface of T cells called PD-1 (Programmed cell death protein 1), and activation of this protein inhibits T cell activity, which is the body's mechanism for regulating T cell activity. This is the body's mechanism for regulating T cell activity. After all, overactivation of T cells can damage other parts of the body as well, so there has to be a mechanism to inhibit their activity, and this mechanism is exploited by tumor cells.
In order to avoid being attacked by T cells, tumor cells produce PD - L1 (Programmed cell death 1 ligand 1), a protein on their cell surface. PD-L1 binds to PD-1 on the surface of T cells and inhibits T cell activity, allowing tumor cells to avoid T cell attack.
To counteract the ability of tumor cells to inhibit T cells, scientists have developed blockers of PD-1 and PD-L1 that do not bind to each other, allowing T cells to remain active. PD-1 / PD-L1 blockers are now FDA-approved immunotherapies, but unfortunately, not all cancer patients are eligible for this therapy, so why?
The role of PD-1 and PD-L1 and the use of their inhibitors (Anti PD-1 / PD-L1). Immunotherapy: Anti PD-1 and Anti PD-L1As mentioned earlier, tumor cells inhibit T-cell activity with PD-L1. But if a patient's tumor cells don't express PD-L1 today, then giving the patient a PD-1 / PD-L1 blocker won't do much good.
In addition, the tumor tissue will form a complex tumor microenvironment in the bodyNote 2, in which multiple cells will have complex interactions, which makes it difficult for T cells to reach tumor cells in the microenvironment. Even if you give PD-1/PD-L1 blockers, if the T-cells can't reach the tumor cells, it's still useless.
Therefore, a way to quickly determine whether a PD-1/PD-L1 blocker is effective against a tumor is very important for patient care. This not only saves healthcare dollars, but also allows patients to switch to other effective treatments earlier, increasing their chances of survival.
Immunohistochemical Staining and Its ChallengesOne of the ways that the FDA currently recognizes the effectiveness of PD-1 / PD-L1 blockers is by using immunohistochemical staining of the patient's tumor tissue ("immunohistochemistry"). Immunohistochemistry (IHC)Note3 is a method that allows specific proteins to be visualized exclusively by staining in tissue sections. Thus, by IHC, it is possible to determine whether the tumor tissue exhibits PD-L1, and also to observe the interaction of T cells with tumor cells after administering PD-1/PD-L1 blockers to the patient.2
The schematic diagram (above) and the actual T cell-tumor cell interactions used to determine whether T cells interact with tumor cells are shown in the following table. Schematic diagram (top) and actual IHC diagram (bottom).IHC may seem like a good way to determine whether a T cell is interacting with a tumor cell, but it has a number of limitations.
First, the IHC needs to be stained, and the fluorescent stain is commonly used today. The signal generated by the fluorochrome dye is strong enough to allow researchers to determine whether a protein is behaving. But as the number of fluorochromes used increases, these signals can interfere with each other.
Additionally, when researchers zoom in on a tissue sample, the resolution of the fluorescent signals decreases, making it difficult to determine how much of the protein is being expressed and how the cells are interacting with each other. Next, there is the issue of signal presentation, where researchers usually try to collect as many fluorescent signals as possible in the same plane so that they can get a clear view and compare signal strengths and weaknesses. However, the tissue sections themselves are three-dimensional, so if you only select the fluorescent signals from the same plane, you will miss the cellular interactions in the other dimensions.
And when the three problems mentioned above (interference from multiple fluorescent overlays, reduced resolution of the fluorescent signal after magnification, and fluorescent signals in three dimensions) are applied to tissue sections with millions of cells, the problem becomes even more intractable. Such image data is a huge project just to compile, and even more difficult to interpret in depth.
Results of multicolor fluorochrome staining in tumor tissue sections. Figure/Overview of Multiplex Immunohistochemistry Astronomy meets Tumor Biology - the birth of AstroPathWhile these problems are a big challenge for biologists, as mentioned at the beginning, they are not hard for astronomers to solve. It's not a problem for astronomers. So in 2018, Taube and Szalay, two experts in different fields, hit it off and set out to create a model that could analyze images of multifactorial tissue slices, based on astronomy's image-processing tools and methods. And the following year, at the NIH's Data Science Symposium Series, they talked about how the technique of depicting galaxies could be used to map the microenvironment of tumors, in the hopes of understanding their structure and vulnerabilities.3
Video of Taube and Szalay's presentationIn 2020, Johns Hopkins collaborated with The Mark Foundation for Cancer Research to create a model that could be used to analyze multifactorial tissue slices, which were then used to create a model of the tumor's microenvironment. Foundation for Cancer Research to create a new Cancer Research Center. The Center brings together experts in the fields of astronomical image analysis, pathology, computer science, cancer genomics, and immunology to build a platform for analyzing pathology images using astronomical methods - AstroPath4.
In a study published in Science in June of this year,1 the team showed how AstroPath can be used to transform multifluorescent-stained images of immune tissue sections into a single, multicolor image with a resolution that allows for the interaction of individual cells.
The complete multifluorescent immunostained tissue section image produced by AstroPath has a resolution at the level of a single cell, even when zoomed in. Reference 1