From Biomarker Strategy to Spatial Proteomic Panel: Designing Custom Panels on Orion™

Welcome everyone and thank you for joining us today. I am Robert Wilson with RareCyte and the presentation is “From Biomarker Strategy to Spatial Proteomic Panel: Designing Custom Panels on Orion™”.

In this 30-minute practical session, we will demonstrate how to translate a biological research question into a custom spatial biomarker panel on Orion. The session introduces the Orion Panel Designer and shows how it configures a panel and assigns each biomarker to its optimal channel.

As a note, we will have a short Q&A at the end of this session. Please submit your questions through the Q&A box and we'll try to get through any questions submitted depending on time.

I would like to introduce our speaker, Mickael Meyrand, field application scientist with RareCyte based in Europe. Mickael.

Mickael

All right. Thank you, Rob. And thank you all for joining us today.

So the question I'd like to address is a very practical one. How do you go from a biological question to a focused spatial proteomics panel that you can trust on precious samples? So we will spend a good part of this session live in the Orion Panel Designer using an ADC trial in breast cancer as our example.

So here is a quick roadmap for today. First, a few principles of the panel design on Orion. Then the core of the session, a live demo in the Orion Panel Designer. And after that we will look at multiplex images obtained with this ADC panel for breast cancer.

And Rob will read out your questions at the end, so please feel free to post them in the Q&A box at any time.

So to answer a translational or clinical question, like the breast cancer ADC trial that we will use as an example today, you need two things – enough plex to measure the cell populations that matter and enough throughput to reach statistical power across a cohort. Historically, those 2 requirements have pulled in opposite directions: plex or capacity. Typically, cyclic immunofluorescence gives you a lot of information. But it is throughput limited when you need statistical power. On the other side, IHC or low-plex multiplex IF have the capacity, but they measure too few markers for this kind of question.

Orion fills that gap, 20 channels in one scan, and 20 samples per day. That's the essential information for large-scale studies, whether you are looking for biomarkers of therapeutic response or for predictive and prognostic biomarkers.

First, how do we build a panel on Orion? So Orion enables high-plex staining and imaging in a single round. There is no cycling, so it's fast. The tissue is preserved, and there is no staining order to optimize.

Second, the staining uses a panel of antibodies, each directly conjugated to an ArgoFluor dye, so you can pick them from our catalog. More than 600 conjugates, all IHC-validated, and all working under the same antigen retrieval condition.

For antibodies that are not in our catalog yet, you can conjugate them yourself with our ArgoFluor kits, or our services team can do it for you.

And third, designing the panel means assigning up to 18 biomarkers to mutually exclusive channels. Orion reads 20 channels: one for DNA, one for tissue autofluorescence, and 18 for biomarkers. The Panel Designer does this assignment for you, as we will see in the demo.

Orion Panel Design Strategy

So there are three principles that guide every panel design decision.

One, pick the biomarkers that are essential for the study. Each one should answer part of the question: a phenotype, a functional state, a tissue structure, or the therapeutic target. Channels are a limited resource, so you use them for the markers that you really need rather than for nice to have markers.

Two, match the expression level to the channel sensitivity. On the right, you see the 20 Orion channels with their sensitivity rank. Rank 4 channels are the most sensitive, so they are the right place for the low abundance targets. Rank 1 channels are best for very abundant markers. And if you can, keep a sensitive channel open, it makes it easy to add a marker later without redesigning the whole panel.

Three, the co-localization: Markers expressed in the same cells, and in the same cell compartment, they share the same pixels. So, place them in spectrally distinct channels. This limits pixel saturation and makes it easier for the software to automatically extract the signal of each individual biomarker.

Every panel should start with a question, not with a catalog. So today, our example is an antibody drug conjugate trial, an ADC trial, in breast cancer, and the biological question is: what are the essential biomarkers to track response to ADC therapy? And this question breaks down into three parts.

First, the ADC target expression: which tumor cells express HER2, Trop-2, and/or Nectin-4, and at what level? Expression varies within each tumor and from one patient to the next, so it has to be measured cell by cell.[MM1.1]

Second, the state of the tumor. Is it proliferating? And what is the status of the estrogen and progesterone receptors, ER and PR, which, together with HER2, define the breast cancer subtype?

And third, the state of the microenvironment: which immune and stromal cells surround the target positive tumor cells? So we want to track T cells, cytotoxic or exhausted, macrophages, including M2-like ones, blood vessels and the stroma. And each of these populations maps to a few precise biomarkers, and that will lead us to our panel.

Building a Panel

So we will build the panel in two scenarios. In scenario one, the ADC target is HER2, which is already in the Orion catalog. So we will build a breast tissue profiling panel exclusively from the catalog. This will show you the main steps of the Panel Designer.

In scenario two, the program moves beyond HER2. It moves to Trop-2 and Nectin-4. Trop-2 ADCs are already approved in breast cancer, while Nectin-4 is still investigational. Neither is in the Orion catalog yet, so we will add them to the same single round panel as custom biomarkers.

So let's start with scenario one. We will create the panel, add the biomarkers, declare co-localization, set a few constraints on clones and channels, and let the algorithm assign the channels.

So from our website, rarecyte.com, click on the top right to access the customer portal, enter your credentials, and log in. And here you have access to documentation, support, the reagent catalogs and the Orion Panel Designer.

This page lists your panels. The green ones are complete and the orange ones are still in development. You also have access to the RareCyte panels. Many pre-configured panels, human and mouse, across a wide range of indications, each with its list of biomarkers.

And here are the panels that your collaborators have shared with you.

Back to my panels, I click here to create a new panel, I give it a name, webinar-breast. It's a human panel for Orion, validation tissue breast, project webinar, lab RareCyte, and I click create. The biomarker list is empty for now, and I can fill it from the themes.

So you can see the list and those are our most ordered markers. I can also add markers directly from the library, which takes me to the catalog. I type CD3, the T cell marker. I click on add and it goes straight into the biomarker list.

So to save time today, I will start from a pre-configured panel, the Tissue Profiling Core, with 10 biomarkers. CD3 is already in my list. So I add all the others. So we now have 10 biomarkers.

I also want Granzyme B, a marker of cytotoxic activity, so I go to the State theme and here it is. I click on it, it's added. Since we are building a breast cancer panel, I add the estrogen receptor ER, and HER2, the human epidermal growth factor receptor 2, and the progesterone receptor PR. So we now have a 14-plex panel, I click on save, next.

Co-localization

So co-localization now. Markers that share the same cells and the same cell compartment and therefore the same pixels must not sit in channels that crosstalk. In T cells, we have CD3, CD8, and PD-1 at the membrane, and Granzyme-B in the cytotoxic granules. In macrophages, CD68 is in the lysosomes and CD163 at the membrane. In tumor cells, ER, PR and Ki-67 are all in the nucleus, while HER2 and Pan-cytokeratin that tends to overlap at the cell edge. So I click save, and then next step.

So, this page lets you link biomarkers to channels and lock channels. Everything in blue is a conjugate already available in the catalog. The channels fall into four sensitivity categories, max in red, high in gold, medium in blue, and low in black.

And so now I would like to exclude some clones. For CD3, we have two clones and I keep only the first one. Same for CD8. And for CD20, I actually prefer the direct conjugate to the digoxigenin (DIG) version.

So to leave room for the panel expansion, I keep two max sensitivity channels, 676 and 724. I keep them free for new low-abundance targets. And so I exclude them for now so that no conjugate is assigned to them. And I can also force the system to use a specific conjugate, for example, one that I already have in my lab fridge like SMA on this ArgoFluor 660L. I press save and next.

And the system now calculates the best panels based on the constraints that we have set.

Review Matrix
So here we have five proposals, panels 1 to 5, and as you can see, they are slightly different from each other. In orange, the spatially co-localized markers. There is no crosstalk at all between them. And in black, crosstalk values above 0.25, and those markers, they are not co-localizing. And so Orion automated spectral extraction will handle them perfectly. So I select, I'm going to select panel 1. And so one last review of the biomarkers, clones and fluors.

And at any time, I can go back to any step to adjust a little bit the different information.

Export and quote

Once I'm happy with the panel, I switch it to Complete and I can export it to CSV. I can share it with a colleague, with an email and a message. No need to have any accounts, and they will receive the panel. And I can also request a quote.

Here on the quotation page, I can adjust quantity and size, 15 or 50 microliters. 50 microliters are enough to stain at least 30 whole slides. You can enter your name, company, email, phone, any comment, for example if you need extra items such as conjugation kits or mounting medium, and you just click submit. You can review your quote request at any time, and you will receive your quote directly by email.

Scenario two

So now scenario two, we need to incorporate two new ADC targets, Trop-2 and Nectin-4, which are not in the Orion catalog yet. So first things first, Trop-2 and Nectin-4, they go through our custom biomarker process in four steps.

One, we validate the clone by IHC on a positive control tissue.

Two, we identify the best Orion channel for each marker, and for that we run a brightness screen with indirect staining, which tells us how bright the target is, and therefore which ArgoFluor dyes are sensitive enough. Trop-2 is a bright biomarker, so many dyes qualify, and our validation work placed it in ArgoFluor 548, which is a standard-sensitivity channel, which is ideal for bright targets. Nectin-4 is much weaker, and our validation work placed it in ArgoFluor 572, one of our most sensitive channels.

Three, we conjugate the clone directly to the selected dye with a simple amine chemistry that anyone can do in their own lab, and we confirm the staining against IHC, as you will see on the next two slides.

And four, you add the new conjugate to your library so you can use it in the panel designer like any other catalog conjugate.

Here is Trop-2 on breast cancer, on the left, chromogenic IHC, and on the right, the single-plex IF on the Orion, with the direct ArgoFluor 548 conjugate, which is in yellow, and Hoechst for the DNA is in blue. Same membrane pattern, same heterogeneity between tumor nests, the conjugate is ready to join the panel.

By the way, Trop-2 is expressed at the membrane of tumor cells, so in the Panel Designer, we will declare that Trop-2 co-localizes with HER2 and pan-cytokeratin.

And here is Nectin-4, the IHC signal is weaker than for Trop-2, as you can see, which is exactly why our brightness screen pointed to a high-sensitivity channel, ArgoFluor 572. And again, the direct IF on the right, which is in yellow, reproduces the IHC membrane pattern. So like Trop-2, Nectin-4 is expressed at the membrane of tumor cells, so we will declare that it co-localizes with HER2, pan-cytokeratin, and Trop-2.
And so now we are back to the online Orion panel tool.

So for the second scenario, I'm going to clone the panel project from scenario one. I can name it webinar-breast-ADC. And cloning creates the new panel project with the biomarker list from scenario one.

So if I go now to the library, you can see that Trop-2 is not in the catalog yet. So I have to go to My Account, My Reagents, and I click the plus button to create a custom biomarker.

Biomarker name: Trop-2. Clone: EPR20043. Validation tissue, going to be: breast cancer. Species: human. It's a direct conjugate, so I can enter the ArgoFluor tag 548. We validated it at 200X, but I could also go to concentration. I assign it to the ADC theme so it's easier to find it later. Same story for Nectin-4, biomarker name, validation tissue, species, and here the tag is one of our most sensitive channels, which is ArgoFluor 572, the staining concentration is 200X, and the theme is also ADC.

And so now back to the Orion Panel Designer at step one where we add biomarkers. And you will see I can simply grab those two new biomarkers, either from the library, so if I type now, if I go to the library, type Trop-2. It's now part of the catalog, and I can add it to the list or from the themes: under My Themes, the ADC theme contains Trop-2 and Nectin-4. So now we have 16 biomarkers in this panel list.

So we could go up to 18, but I actually prefer to refocus on a smaller panel by removing the B cell marker CD20 and removing Granzyme B, which are not really central to the question for this trial.

So we are back to 14 biomarkers. From here, the workflow is the same as scenario one, co-localization, channel assignment, and quote.

The final panel

So let's go straight to the panel that we actually stained. Here is the panel that we actually validated and stained on breast cancer tissue for this webinar, a 15-plex with 14 biomarkers plus Hoechst for DNA, all in a single round with direct conjugates. On this table, the colors group the markers by role. You get the tissue architecture, the tumor, the ADC targets, immune cells and cell states. The 3 ADC targets are in gold, Trop-2, Nectin-4, HER2.

You can note that we have four channels that are still open, including two of the most sensitive ones, so the panel can actually still grow as the program evolves.

Before we look at the images, a quick word on the path from design to study. First, develop. Once the panel is ordered, you titrate it on the relevant tissue. This is also the time for you to define your analysis strategy. Second, validate. Do a reproducibility study on a few pilot samples. And third the clinical sample testing. You can stain up to 24 slides per batch in a single round of staining and imaging and the Orion HT scans 24/7 with 30-slide automation. Along the way, Orion generates a QC report automatically for every single scan, and the quantitative analysis runs automatically down to the single cell data, cell population assignment and density per region of interest.

Stained slides are stable for more than a year, so you can image the slides whenever you want. You can also run an H&E on the same slide after imaging since the Orion also has the brightfield capability. And if you need to investigate some samples with more markers, you can even add extra immunofluorescence rounds on the same slide, even months after the initial staining.

And now the images

This is a whole slide view of the breast cancer specimen stained with the full panel. A few channels are shown here, CD163, SMA, Trop-2, HER2 and CD31. We also get the tissue autofluorescence. At higher magnification, the three ADC targets, you have HER2 in blue, Nectin-4 in red, and Trop-2 in green. You can see varying levels of all three ADC targets on individual tumor cells. This is exactly the heterogeneity question that we started with, and now it can be quantified cell by cell.

Let's zoom into our first region. The tumor cells express varying levels of Nectin-4, which is in blue, and Trop-2 in green, and many of them are proliferating because you can see in magenta, the Ki-67 staining. The macrophages are in red, they surround and infiltrate the tumor, while CD8 T cells, which are in white, they stay at the periphery.

Same region, but different markers. All the tumor cells, they express HER2, which is in white. While you still get a subset that also expresses ER, PR, or both. The T cells, they also surround the tumor, while macrophages, they surround it and infiltrate it.

And we still stay in the same region. Here, the tumor vasculature: tumor cells (pan-cytokeratin, in white. They are surrounded by blood vessels with CD31 on the endothelial cells in magenta and SMA in the vessel wall, which is in cyan. You get a subset of T cells in green. This subset is also exhausted since they are PD-1 positive, in red.

And now we get a second region. Here, most tumor cells express Trop-2, and only a few of them express Nectin-4. So the balance between the two targets is different from the first region. Some tumor cells are proliferating. Macrophages, they surround and infiltrate the tumor, while CD8 T cells, they stay at the periphery. Same region with the same breast cancer subtyping markers as before.

Finally, last image, the tumor-stroma interface. Large tumor islands are surrounded by a layer of SMA-positive cells in cyan, which keeps most of the T cells, which are in green, the T cells outside kind of form of immune exclusion. And some of those T cells, they are also exhausted. Please keep in mind that all of those readouts, they come from one single slide, stained and imaged in a single round.

Key takeaways

To wrap up, four takeaways.

1. Always start with a question. Define the cell populations and their states first, then choose the biomarkers that resolve them.

2. Design for performance: you must declare the co-localization, match the expression to channel sensitivity, and if possible, keep channels open for the future. High-performance panels, they are really the foundation for trustworthy data.

3. Mix catalog and custom reagents freely since they all are sharing one antigen retrieval condition.

4. And because Orion works in a single round, there is no staining order to optimize. The Panel Designer assigns every biomarker to its channel.

Thank you very much for your attention. And Rob, over to you for the questions.

Q&A

Robert

All right, thank you, Mickael. Yes, we got a couple that came through and again, anybody else who would like to add to that, please send them over through the Q&&A.

Mickael, the first question, “you mentioned that I need to titrate the panel once I receive it. How exactly is that done?”

Mickael

Okay, so for panel titration[MM2.1], for a full panel, I will recommend to do two serial sections of your tissue of interest. The full panel plus Hoechst at 200X on one slide and 800X on the other, and for each marker, you can keep the dilution, or a titer in between, that gives a strong signal with the expected pattern, low background and no camera saturation. So the camera is 16-bit, it's about 65,000 maximum. So we tend to stay below 30,000.

For your own conjugates, you stain your tissue of interest or a control tissue that expresses the target. Usually, I recommend to test 0.1 microgram per mL, 0.3 and 1 microgram per mL. You compare it to IHC, and then you keep the best concentration with the same criteria that I mentioned. And then, of course, you revalidate the multiplex at the selected titers on a couple of samples, and you can go and run the cohort.

Robert

All right, fantastic. Okay, another one. “What is the biopsy criteria? Needle gauge size, tumor volume, passing criteria, etc, to be able to obtain the data as shared on the screen for the ADC application?”

Mickael

So that's a very general statement, but you know on the Orion, everything that you can put on a standard glass slide, a microscopic glass slide, even if it's FFPE or fresh frozen, cut somewhere between 3 micrometers and 10 micrometers, we can image it. As I used to say, if you are able to get something on IHC, then it will work on the Orion. There is no doubt. We always validate everything against IHC. So the first step is to actually get something with IHC, and then you are fine to move to the more complex high plex on the Orion.

I hope that answered the question. Otherwise, you know, feel free to email us and we can come back to you.

Robert

Okay, thank you. “For custom biomarkers, how exactly do I know which channel to put them in?” I know you covered that a little bit.

Mickael

So, for custom biomarker, so, actually, the best channel will depend on the expression level of your target. So as I said, we validate the clone by IHC on a positive control tissue. So usually the IHC signal often already suggests the channel category. But then to be more precise, we run a kind of brightness screen on the same tissue. So we take the unlabeled clone, we reveal it with an anti-mouse or anti-rabbit secondary using one of the most sensitive channels. And we do a specific type of imaging, it only takes a minute, but we image with three exposure times, and usually the longest exposure without saturation gives the brightness, and our dye selector lists the ArgoFluor dyes that are sensitive enough.

So if I take the example of today, there was like 10 channels for Trop-2 that qualify, but for Nectin-4, there were only 2 or 3 channels that could be used.

So we pick one, keeping the most sensitive channels for the targets that really need them. Or we can let the Panel Designer choose. So in that case, you go to the tool, you can allow all the qualified dyes for your new target. And then you declare the co-localization and the algorithm will pick the best channel the rest of the panel.

So of course, then you have to conjugate, compare to the IHC, and titrate on the tissue of interest. So it's a long answer but it's actually faster to do than to explain the different steps.

Robert

All right, fantastic. Now, I know we planned a session for 30 minutes, and it looks like we're going a minute or two over. I still have a couple other questions, but at this time, I'm going to make sure that our team is able to follow up with each of those questions directly. Certainly, anybody who wishes to contact Mickael or our team, support@rarecyte.com and our team will be happy to get back to you fairly quickly to talk with you about those.

And I'm sending over a quick poll for anybody that's interested in getting a direct consultation meeting one-on-one with Mickael or somebody from our other teams. If you're interested in that, please let us know.

Otherwise, thank you everybody for joining us today. Have a wonderful day, wonderful evening.

Thank you.