Executive Interview with Suzanne Coberly

[0:03] Good afternoon, everyone, and welcome back to another Oxford global interview. Today. I'm joined by Suzanne Coberly, Senior Director of Translational Pathology at Bristol Myers Squibb. Suzanne leads a discovery and translational pathology lab in San Diego supporting BMS oncology and immuno-oncology programs, working across many phases to develop predictive assays and biomarkers that guide precision medicine. And today we'll be discussing the challenges and applications of digital pathology in pharma. So thank you very much Suzanne for joining us today. And yeah, I guess the first question from me is, could you maybe just tell the audience a little bit more about your current role and the purpose of the translational pathology lab that you lead?

Yes, I'm glad to do that. My lab is actually a research pathology lab. It involves both discovery and translational projects, and we are supporting the earlier research in oncology.

[1:07] Bristol Myers Squibb has a number of different groups that do oncology. We're supporting a couple of them, including targeted protein degradation as well as cancer immunotherapy, and so our group supports them from early discovery through early clinical and then after that, it's transferred to another pathology group. We do a lot of combination of animal studies, target validation and biomarker studies, where we're looking for predictive biomarkers.

[1:39] That's great. Thank you very much. And I guess you mentioned a bit about bridging the gap, and what kind of excites you the most about leading a lab like this?

[1:49] I actually am very excited by helping finding the data to bridge the gap, as you say, between discovery and clinical what I like to do is to do the biomarker work early on, try to figure out which patient populations might be responsive to this particular drug, and try to get a idea of efficacy versus, you know, either expression or other types of pattern recognition. So then and then, what I like to do is to see that assay transferred into the clinic and used in the early clinical trials, and hopefully we may get a predictive biomarker assay.

[2:25] If not, we at least get assays that we can use in the clinic for the drug development, looking at PD or mechanism of action, or other types of other types of things that are useful to get the drug developed.

[2:40] That's great. Thank you very much. And to what extent are you guys using spatial biology approaches, and is there a way that you ensure that these are well integrated with existing pathology and image analysis pipelines?

[2:58] Well, if you're talking about, if you're talking about the RNA spatial analysis, like 10 times genomics, we don't personally use it, but we do work with the groups that use it. I do want to say, though, that I feel that standard pathology is spatial biology. A lot of people have come up with this new name as though it's something brand new. And while the technologies are new, the pathology always has been spatial biology. As a matter of fact, even something like IHC, it's most useful in cases where you need to look across the tumour and look at heterogeneity or other types of spatial biology questions. And obviously multiplexing is something that you really do want to use to try to get an idea of cell to cell interaction on a protein level. The newer technologies, with the RNA and the 10 times genomics are really good, and they help extend that, but they're not really anything different from a strategic point of view. They're more something that's hitting some of the areas that standard pathology can't get. And so what we tend to do is we tend to think about what we need and what technologies will work the best for that. And we also have to balance that with practicalities of doing work in the clinic, with, without a lot of tissue, and with, sometimes, you know, the newer technologies. And by newer, I mean almost anything that's not just standard IHC, because, you know, in the clinical labs, a lot of those things aren't used very often, but we need to balance the difficulty of doing that with the what it brings to the table. And sometimes what we'll do is we'll do the newer technologies and help out our people who are developing the clinical trial biomarker plan to do that. So that's sort of where I'm at the moment.

[4:44] Okay, perfect. Thank you so much. And when, in your opinion, when you think about digital pathology in pharma, what would you say are the top priorities right now?

[4:57] The top priorities? I'm not quite sure what you mean by that.

[5:01] the top opportunities, sorry, ah, well, I mean, honestly, almost everybody I've worked for in the last 10 years uses digital pathology, because in research, particularly in pharma and biotech, I don't know if you could actually do pathology without it. It really leads to, you know, it allows you to collaborate, say, with CROs across the world, as well as just allowing you to do the image analysis that you need to do for certain projects these days. Obviously, we're also using the digital pathology to get into sort of AI right now, I'm interested in something that's virtual IHC, where we can predict IHC or multiplex from the H and E slide. So digital pathology is really vital. Having said that, the one thing that we do have to be careful of is things like different scanners.

[5:55] Well, different stains, of course, but different scanners and how they affect the slides and whether they're darker, lighter, a little less resolution. You have to understand what people are scanning and what their scanner is and compared to what you have. Because when I'm doing IHC, I'm very carefully looking at intensity as well as just that positive, negative, and that can be really different depending on which vendor scanner they're using. So that's one challenge, just to make sure, even across an organization, you have the same scanner, or at least you have scanners that are comparable and that the settings are the same.

[6:33] Thank you very much. And how does digital pathology contribute to patient stratification and by market development, specifically in the immune-oncology space.

[6:48] Well, it's a combination of doing some sort of multiplex analysis in most cases, because, you know, especially with immuno oncology, you can't really do serial IHC sections. I mean, you can, but you can't really. Then look at individual cells and if they have multiple markers, and are they next to another cell that has multiple markers? So first you have to start with a good scanner, and you have, I mean, with a good slide that has actually at least the lower level, like seven to eight markers on it.

Otherwise you're not going to get enough markers to do any good to help you predict anything. And so you have to initially, we start this really early, you know, in the middle of sort of, not really early discovery, but sort of getting it's sort of like the beginning of translational when you've actually had some data on efficacy versus expression. And so then we talked to the biomarker team, figure out what they have, what questions they have, and how we can best answer it.

 And then you then obviously have to go into the whole creating the multiplex the analysis, and you have to do the scanner and make sure you have the right scanner for this that will allow you to then do the analysis that you need. So you have to kind of plan the whole process, the wet lab, the image analysis, or other AI analysis, and then, you know, just the technical scanning to make sure the scanner can help you get the information you need from that slide in order to do the analysis, needs to be set up together.

So the point is, is that digital pathology is really good at allowing people to then analyse the data, but it's also really good to allow us to work with clinical sites. So as I said, you have to have the multiplex. You have to make sure the scanner is good, and then you have to make sure that whoever is doing the assay for you can then share the results and in a way that you can interpret.

So there's a lot of steps along the way for digital pathology. Obviously they couldn't. It would be very difficult to do without digital pathology, because, like, if we have some patients, if we have clinical trials in other countries, for example, it's very otherwise, we would have to get the slides. We'd have to try to look at them. And, you know, obviously, with Multiplex, you really, as a pathologist, I can look at them, I can tell you that the stains are on approximately the right slides, but I can't analyse them.

So you do need the digital image in order to do the analysis. Plus then you can share the information across the world, not just, you know, by using FedEx and shipping slides back to you overnight. So digital pathology, really, it's still the two things, you know, creating a digital image that you can analyse, and also allowing collaboration across many miles and doing it a lot faster.

[9:38] Thank you. And would you say then, so that, like ensuring this kind of seamless workflow, is that one of the main bottlenecks when introducing pathology, digital pathology, and kind of bridging that gap between the discovery and translational settings? Do you think this is the main challenge?

[9:57] If you mean with clinical trials, it's not as big a deal because we work with CROs that have those technologies. But what we do need to be aware of is, can we like, for example, if you're going to do immuno oncology, you do want to multiplex, you do want, usually, immunofluorescence, and so you do need to work with somebody who knows how to do those things, has the right scanner, has the right analysis plan, and so that is a challenge to find the right people.

The bigger challenge, I think, though, is if this turns into sort of a patient stratification biomarker, but then also it may become like, say, a companion diagnostic down the line, I do know that people have been working on getting that into the market. The challenge there then becomes a case of who will do the testing for you? Can you get reimbursement? Because the clinical world really doesn't have as much multiplex as the research world, and so that's probably while we've now made a lot of progress, and we now have more CROs that we can work with in the clinical trials, trying to then go to the clinic could be a very difficult step in that regard.

[11:06]  Okay, I see, thank you. And I have quite a broad question, but from your perspective, what would you say success looks like for digital pathology as a precision medicine tool?

[11:23] Well, I do think that digital pathology, whether it's of Chromogenic slides or of multiplex immunofluorescence slides, I really think success would be getting the vendors to kind of get together with maybe the users, and to have a common some sort of common ground. In other words, they're doing it to some degree now, but different scanners are very different sometimes. And so what we need to do is have some sort of common like, what makes a good scanner? We need to agree upon that, and we need to get that done.

And so it's that is something that's somewhat difficult, maybe not as much for the Chromogenic as for the Multiplex, because, because that may depend upon which technique you use to do the Multiplex, but I still think people do need to get together, and we do need to have some advisory talks on this, and to Have everybody agree, so that maybe vendors will follow a certain standard. Right now, they're not really.

The FDA did give people permission to use scanners to do reading of slides in the clinic, and that actually didn't occur till covid. And so it was really kind of amazing that we we've had these scanners around for years, I know, at least for 10 years, and I've used them for that long, they're very good, especially now the fact that people who were pathologists who were trying to read slides for diagnosis couldn't use them at all, and except for maybe at their desk in their office, which kind of defeats the purpose. So the bottom line is, let's try to get a good scanner that actually is pretty consistent between vendors. Let's have some recommendations for that.

And if we can get that consistently and get it done, I think that would really help a lot, because the whole analysis, you know, we're trying to do more analysis, more AI, but we really want to have a you know, if you have slight scan at one place and they look one way and they're scanned another place and look another way, your analysis won't be the same. And so you really need to have some sort of consistency. I think that's where it comes down to.

[13:38] Thank you very much. Yeah, I think the emphasis on the standardisation process is very, very essential. And I guess, okay, I final question for me, from me Suzanne, is, obviously, we're, we're very excited to be welcoming you at Oxford Global's biomarkers and precision medicine event, and you'll be participating in a panel discussion on digital pathology and pharma overcoming integration strategies and future proofing. Are there any particular challenges or integration issues that you would like to highlight in the panel that you think the audience would relate to?

[14:20] Well, one thing, like I said, it's the whole standardization, making sure that whoever you're working with, like, for example, we're working with somebody in China, and we can't have certain samples shipped back. So what we need to do is we need to give them our protocol so they can stain the slides there, scan them, and then we can look at them. The problem with that is, then they have a different scanner than we do, so we have to give them some controls that are already stained, because they can't even ship back the stained slides. So it's, it becomes a little bit of a it's something you have to look at the workflow and how you can do it. Because just because you stand a slide, you scan it, and you um, you know, have somebody look at it. You have to be very, very careful that your scan matches reality, and sometimes they don't. So there's that.

There's also the other challenge that is just something to think about. It's completely different, is storage, because these digital slides are extremely huge. And so while I know that storage is pretty cheap these days. You still have to figure out, can you afford to keep all of your slides, and if you're going to get rid of some of them after a while, how do you do that? You have to have sort of a workflow in that regard as well.

And so that those are a couple of things. I know they don't sound really sexy, but if you don't get the basics right. You're not going to get good data out of it. And so that's, that's the two that I think are the biggest. The other one is, of course, I really am into Multiplex, because I feel that for immune-oncology, pathology needs to go well, not exactly like genomics, because genomics can do hundreds and hundreds of slides, but we do need to go towards the multiplex. We do need to make that more standardized, make it a little more easy to use, and then do the analysis. And digital pathology is part of that. And so we need to make sure, because right now we're working with concentric, and concentric is like a image management system. They are getting there, but they don't even have the ability, right now, to store multiplex immunofluorescence Slides and show them to us in a way that the colours are right. And so that's another challenge to think about. You might want to integrate your multiplex with, say, your Chromogenic slides. So those are a few things that you got to get it right, and then you can go do the really fascinating things with AI and image analysis.

[16:45That's great. Thank you very much. And yeah, I think that brings our interview to close. So thank you, Suzanne, for sharing your thoughts on the evolving role of digital pathology, and we very much look forward to welcoming you at our event, thank you. I look forward to being there. Thank you. Bye, bye.