Intro: Welcome to Precision Pathology Podcast, featuring interviews with authors and global thought leaders who are at the forefront of technological advances in precision diagnostics and therapeutics. Your host is Dr. George Netto, the Simon Flexner Professor and Chair of Pathology and the Laboratory Medicine at the University of Pennsylvania's Perelman School of Medicine in Philadelphia. Here is Dr. Netto.

George Netto (Host): Welcome to a new episode of Precision Pathology Podcast, and as I call it P Cubed. My guest today is actually an internal guest. Last time, we had to go all the way to Heidelberg to meet our previous guest. But today, my guest is Dr. Sydney Shaffer, and she's a member of our department. She's an Associate Professor in Pathology and Lab Medicine, but also in Bioengineering. And she's a typical example of recent members of the pathology field tent where we gather all the expertise in technology. And actually, she's a member of a division that was created here at Penn a few years ago by my predecessor. And it's termed nicely Division of Diagnostic Innovation. And this will be the topic that Sydney is going to talk to us about, her expertise in cutting-edge technology that fits very nicely into her interest in cancer and cancer immunobiology, and how to use single-cell techniques and other techniques to understand how cells differentiate how they respond or don't respond to drug therapy, invade or don't invade. So, she's really a delight, one of our rising stars. And believe it or not, as an assistant professor, she's already our associate vice chair for basic research, and that speaks volume about her intellect.

But more importantly, Sydney is one of our very active team members in the department. So, I was delighted to have you here, Sydney.

Sydney Shaffer, MD: Thank you. It's a pleasure to be here today. I'm excited for the discussion.

Host: All right. So, I thought we're going to spend toward the end some time talking about your recent publication in Nature Communication. But I wanted to use this episode at least as a meet-the-expert about cutting-edge technology that's really applicable to pathology and that really will play a major role in precision pathology as the title of the podcast. So, maybe we can start and have just a casual conversation about what kind of technology you see promising and why you're using and spending all our money on these technologies right now.

Sydney Shaffer, MD: My lab has been really excited about spatial transcriptomics technologies. These are technologies where we can go into fully intact tissue sections from people, and we can profile individual molecules of RNA and pick them up and know precisely where they're at in spatial context. And I think this is really beautiful, because it really leverages what pathology knows, which is that spatial context matters for everything in biology and in human tissues. My group has really spent a lot of time and, as you said, money investing in the latest and greatest spatial transcriptomics technologies.

Host: Of course, there's no commercial biases here. So even if you mention the name of the platforms, that's just so people know what we're talking about. Can you tell for audience who really don't-- clearly, a lot of pathologists don't use that on a day-to-day basis, and we'll talk about that at the end, how is this going to ever become applicable, CLIA to CLIA lab and something we use, which is the hope. But currently, what do you see? What are the differences between single-cell versus bulk?-- I guess that's the past-- and the spatial and the advantages that you see and why you are moving away from bulk sequencing and all that.

Sydney Shaffer, MD: Yeah, I'd love to. If we kind of backtrack, I jumped straight to the fanciest and the most complicated, which is spatial. If we backtrack, historically, like sequencing started with bulk sequencing, which is you take a piece of tissue, you blend it all up, you make a smoothie of all of the contents inside it, and then you go in and sequence that. And that really takes apart everything that's in there and mixes it all up into one signal. So, you get one signal from that whole sample.

The kind of first revolution from that was single-cell technologies. And so, the idea there was that instead of blending everything together, we're still going to take a tissue apart. So, you may take a tissue, you may disassociate it into single individual cells, which is tricky, and has all kinds of complications as well. And then, take each of those individual single cells and sequence it by itself. And that's really informative because that teaches us about how heterogeneous a tissue is. It teaches us about what cell types are within it, how different are they from each other?

So, these technologies have really informed, I would say, kind of all fields of biology and medicine tremendously. But as I mentioned at the beginning, the first step of doing single-cell is taking the tissue apart. And what we've been really excited about also are now technologies where you don't have to take the tissue apart, but you can still get that resolution where, you know, down to an individual cell what is different about each cell.

Host: Wow. And clearly, needless to say, you're lucky being in a department, basic scientist in a department of pathology, where you just have to show the slide across and work with your colleagues, anatomic and surgical pathologists or neuropathologists, where besides your knowledge, you're getting their knowledge about spatial and about the topography of the cells, usually on a 2D slide, right? So, mention an example where you feel this spatial context will come to the esophagus-- and Barrett's specifically, the topic of the manuscript. But in general, what have you applied in the context of cancer in your own research some of these technologies?

Sydney Shaffer, MD: Yeah. With spatial transcriptomics, we've been really excited about better characterizing all the components and interactions going on inside of tumors. We've been doing a lot of work more recently in pancreas cancer, which has been very exciting looking at not just what are the cancer cells doing and what is heterogeneous and different about the cancer, but looking at the interacting immune cells and stromal cells and really identifying discreet components of the tumor that are representing cell-cell interactions that we think might be fueling the cancer because we know cancer doesn't act in isolation. It's got an immune compartment and a stromal compartment. We think those are also part of the evolutionary ecosystem that the cancer is building that makes it more aggressive and complex.

I would say that being in a pathology department is really the perfect place to do this kind of work, because a lot of the observations that really stirred us digging into this deeper have come from conversations with my colleagues in the department, one of the things that's really just inspiring and motivating about this type of work.

Host: Wonderful. So, before we jump into the paper, to me at least, you know, and to many pathologists, the data points, some of these technologies taken 20,000 gene or 5,000 even, the ones that are at the subcellular level. It's that amount of data and making it practical, reproducible. You know, pathologists, we are obsessed with reproducibility, calibration, and all these. Because if we're going to bill for this and apply it to patient, we better be very accurate.

How do you see the technology, you know, in the years that you've been involved with? Is it becoming easier to decipher this? Do you see this as one day a lab-- like, right now, we have molecular diagnostic labs that mainly are bulk. You know, unfortunately, you mentioned all the disadvantages of mixing-- what do you call it? A smoothie. But the reality is we take a smoothie of colon cancer, and all that we care about is: is this gene targetable or not? Or lung cancer. So, it's working for us. And we finally got it to a point where we can do it day in and day out. Do you see the same? Is there some hope, some progress? Because a lot of people say, "Forget it, these will never make it to our clinical bedside in pathology."

Sydney Shaffer, MD: Well, I can say that I've seen a lot of progress in the time that I've been doing this in terms of resolution and cost. The technology is getting better and better. And we can measure more and more features, molecules inside of tissues. And with that, it's also getting cheaper as things have moved forward.

I think in terms of this moving into pathology, I think that these technologies right now are at a stage where they can help us identify new markers, limited panels, where we can work them into our clinical framework. So, maybe instead of staining for 20,000 genes in our panel, maybe there's a really smart set of 10 to 20 that would be super informative and that would be a useful platform for a clinical application.

And I think that's how this begins. We start with more limited panels. And as technology advances gets cheaper, our ability to do the computational analysis that these are going to require gets more feasible, then I think it could be reasonable that bigger panels larger sets of genes could be clinically informative. But I think that we've got a ways to go to get there. And I think there's cost and infrastructure and clinical decision-making and many studies that are going to build and pave the path to that final future.

Host: Do you see AI helping us in that interpretation or simplification of the data? You talk in this paper about UMAP and all these things. Do you see it making it, you know, more practical? I guess it's a rhetorical question, but I want to hear it from you.

Sydney Shaffer, MD: Yes. I mean, of course. And I think on the short term, we're already seeing it with a lot of tools that are built around AI-related frameworks that are very helpful in analyses. I think in the future we could imagine things where you would be combining a set of these markers and an AI algorithm And that would tell you your clinically informative information that you need to make the diagnosis. And so, I think as the technology advances, the AI will advance with it. And I think there's a real opportunity there, and we're certainly thinking about it a lot in the lab.

Host: Excellent. One last point when you mentioned-- you know, it crossed my mind-- when you mentioned that, boiling it down, the discovery could be on thousands, but then boiling it down to even 100 or 50 more realistic, the multiplexing with single molecule, FISH, sequential FISH. That's probably more like it. Am I correct to think that that probably will be the earliest thing that comes to a pathologist, you know, in their lab, offering this and with the aid of the new imaging or not necessarily that?

Sydney Shaffer, MD: It's interesting. It's a very good question. I think to back up a little bit, spatial technologies can be divided into two categories. And I think this is what you're referencing. We can divide them into those that are sequencing-based, where we record spatial information and the information we're sequencing or those that are imaging-based.

 I think imaging naturally lends itself nicely to pathology, and it's what we are used to. And we have multiplex immunofluorescence, and other things that look a lot like it. So, I think that it starts there first. I am really interested to see which ends up being more cost-effective in the long run. And I think time and technology will tell. But I think that's also going to be really important, is cost effectiveness and scale up and how we can implement this on reasonable number of samples from patients without it costing thousands of dollars per section.

Host: Just as a point of reference, currently, maybe it got cheaper, but it's not unusual to be five or $10,000 per slide to do some spatial studies, right?

Sydney Shaffer, MD: Yes. That's not unreasonable at all.

Host: Yeah. So, I could tell, being a chair, you know, where all my money is going here in the department, and you're one of the culprits. But the point is that this will get cheaper and you make a good point. Immuno, for example, sometimes we don't think about it, but pathologists all the time. I'm not saying that's correct, but a lot of pathologists order 10, 12 immunohistochemistry and $60 or $100 a pop. That's over a thousand dollars. And now, sequencing and stuff is cheaper. So for the first-time they say no, it's probably better to do a limited panel by sequencing and do the molecular, which wasn't the case-- you know, 10 years ago, we didn't think like that, because it took a long time and more expensive so, I guess we're not limited to the imaging, because that could be also very expensive if we're going to do sequentially and multiple fluorescence and all that. So, who knows?

Sydney Shaffer, MD: Although I will point out that antibodies cost more than oligos. So, immunofluorescence is intrinsically a more expensive test, because it's much more expensive to make an antibody than it is to make a sequence of DNA. So in principle, there is a reality where you could do this with imaging-based tools using DNA probes, and that would be cost-effective.

George Netto (Host): Awesome. So, I guess, how about that? And we will give it five years. We'll bring you back and see which side. Is it the imaging or sequencing-based? And what we're using here at Penn and other places. So, this is really very informative. And I love the conversation so far.

I think the next part's going to be even more focused and more exciting. So, let's talk about your recent paper. And I'll let you feel free to give credit to your co-authors in a second. But the paper basically is Clonal Cell States Link GE Junction Tissue to Metaplasia-- meaning Barrett's-- and Cancer, adenocarcinoma. First, let me give you the platform to give credit, because I know despite that you're the senior author, probably a lot of work was done by Rodrigo, right?

Sydney Shaffer, MD: Yes. The first author of the study is Rodrigo Gier, who was a very talented and hardworking PhD student in the group who is now a postdoctoral scientist.

Host: All right. And like everything, it's team science. The study came out just last month, December 2025. And I want us to start with a little background. Why did you do the study and what's so unique about it?

Sydney Shaffer, MD: That would be great. We were really interested in Barrett's esophagus. Barrett's esophagus, as you mentioned before, is a type of metaplasia. I had been intrigued in following this literature for a very long time. One of the big challenges in the field is knowing who with Barrett's esophagus is going to develop cancer. And this is fundamentally a clinical problem, because there's a lot of people that get Barrett's. And only a really small subset of them are actually going to develop cancer. So if we could predict who's going to develop cancer, it would be very informative, because we would save a lot of time, energy, stress. And we could also intervene on the people that would develop cancer. And so, we were really interested in understanding what is happening in the subset of patients that are more likely to develop cancer.

Host: And of course, the step there is the dysplasia part. And so, basically, you are trying to understand using, again, those cutting-edge technology, and even if it's more cutting-edge as we're going to see in a second, which cell is giving rise to what and where are the clones coming from, and how does these mutations accumulate or don't accumulate. And so, let's start with what technology you used and how did you come about doing all that.

Sydney Shaffer, MD: Yes, that'd be great. So, I told you before about single-cell RNA sequencing where we can disassociate tissues and learn a lot about their molecular state. We're also really interested in understanding how things change through time. And that's a harder problem when we're really just getting one piece of tissue from a patient, because we really just get that one snapshot.

In this paper, we used a technology that falls into this class we call lineage tracing and lineage tracing technologies. The goal of them is to figure out what cells in a tissue came from a common ancestor. Here, you use the word clone. We think of the cells that came from the common ancestor as being the clone.

And so, the way that we do that is actually, I think, pretty cool. So, in our single cells, all of them have DNA. And one of the types of DNA inside of them is mitochondrial DNA. These mitochondrial DNA get mutations at high rates. They have many copies. And so, it's very easy to imagine ways to sequence them.

And so, when the cells are just naturally occurring in the human body, they can acquire mutations in that DNA. And these are like little scars, and they do nothing to the cell. They don't harm it, but they sit there. And when the cell divides, they're passed on. And so, these scars in the DNA can basically be used as a tool to know which cells are coming from the same ancestor.

Host: So like a tag.

Sydney Shaffer, MD: Like a tag, exactly.

Host: And literally when they divide, the mitochondria carrying those scars go to the daughter cells and then to the granddaughter cells.

Sydney Shaffer, MD: Yep. And so, this gives us a way to infer the clones and really to reconstruct like a family tree of a biopsy we take out of a patient. We can go in and ask, what's the relationship between all of this cells and this biopsy? And so, that allows us to really start to look at whether there is a stable set of clones coming from stem cells or whether there's a mean, bad, aggressive clone that is concerning in the patient sample.

Host: Wow. And we'll come to the spatial aspect of this in a second. So basically, to summarize, you're using this, you called it in a previous conversation, like the fourth dimension, the dimension of time, right? So, you're using the scars or changes in the mitochondria. Let's call them bystander right now, because they're not affecting in the oncogenesis, maybe they are. But for the sake of this argument, just to trace those cells and see where that crypt, for example, that we're looking at with dysplasia where these individual cells came from, you can do that.

Sydney Shaffer, MD: Exactly, yes. That's the technology that we were using here in a series of patient samples collected from the endoscopy suite.

Host: So, you know, applied that my understanding to the benign segment, the squamous or the benign, you know, gastric, and then to the transition and you took multiple punch. Hopefully, I'm sure you had an IRB approval for all these things. So, these biopsies and you did it on frozen tissue, I imagine, or not necessarily, or was it paraffin?

Sydney Shaffer, MD: No, fresh. So, we collected them fresh so that we could get the highest quality material possible

Host: And you just dissociated the cells. So, what did you decipher about the lineage? Let's call it lineage tracing of these different compartments.

Sydney Shaffer, MD: Yes. So when you asked first about the benign, when we looked in patients that have non-dysplastic Barrett's or the completely benign Barrett's, we find many individual clones inside the tissue. So, we find many separate little families that are coming from-- you alluded to crypts, they're probably down in these separate glands, each one of them. And they're pretty stable. They're like homeostasis with each other. None are outgrowing one. One may be a little bigger, one may be a little smaller, but it's really kind of a stable tissue ecosystem that we're looking at.

Host: Even within the same cell type, like, let's say the intestinal or, you know, metaplastic intestinal. So, you see families of these intestinal...

Sydney Shaffer, MD: Little families of these intestinal pockets. The other thing that we found in the benign is that sometimes when we look in these families, we find families that also contain the neighboring cell types. So, we're looking at a disease of the gastroesophageal junction generally. And so, the above it is the esophagus, and below it is the stomach.

And so, occasionally, we would find clones that also contained neighboring normal esophagus and neighboring normal stomach, which tells us something about where this comes from because it tells us that Barrett's esophagus must be able to come from both, a progenitor that gave rise to the esophagus, and it also can come from a progenitor that presumably was derived from the stomach. So, it points to multiple tissues of origin.

Host: So, progenitor that was going to head towards squamous and progenitor that was going to head toward benign gastric mucosa cardia. Interesting. So, what happened when you looked at the dysplasia?

Sydney Shaffer, MD: So, the dysplasia is when we really started to see something that broke from kind of this model of a stable ecosystem. When we started sampling patients with dysplasia, what we were finding were large clones. Large clones that had a very distinct and aggressive molecular profile that looked highly different from the normal stable ecosystems that we'd seen in the non-dysplastic disease. And in individual patients, we could see clones that had just really taken over the tissue and were clearly basically outcompeting all of their neighboring stable pockets and are really taking over.

And we think backing up that this really represents kind of the first step of the early cancer evolution in these patients, because they now have made something that is fit enough to start to outcompete and drive kind of progression of the tissue.

Host: And if you were to sample different areas of the dysplasia, these clones can be interrelated or not necessarily, the ones that took over?

Sydney Shaffer, MD: Yeah. So, it can be both. We have seen independent clones, separate ones. We've also seen big ones that we can pick up in multiple locations. We are finding patient to patient differences in that.

Host: And in what did you have cases where you had the whole cancer already progressed to invasive adenocarcinoma, or no, you didn't?

Sydney Shaffer, MD: Not in this manuscript, but that is coming soon. It's in the works.

George Netto (Host): Excellent. So, we'll have to have another podcast about that. So, this is all cool, especially pathologists, usually, we're very interested in oncogenesis and understanding. But I think you touched upon that earlier, The idea of can we intercept or can we identify who's going to go into dysplasia and, hopefully next, who's going to go into invasive cancer? And can we stop it? Do you think any of the information you gathered here can help us toward that proposal?

Sydney Shaffer, MD: So, there's two ways that I think about this. The first is that when we find these early aggressive clones, we now have a series of, A, molecular markers that we can start to test. And I'm working on that with some of our path colleagues. We also have a set of pathways that we could think about targeting in these patients before they even develop cancer. So, this would be like cancer interception strategies. And we're also working on those approaches as well.

The other thing that we have found from this is that not every patient's dysplasia is really the same. And so, we're also starting to scale this up and look across more patients, because we think there are many different flavors of dysplasia that we're going to see in patients. And we see some that are stereotyped. They don't perfectly correlate with the genetics. So, we're really trying to understand kind of the molecular heterogeneity and dysplastic evolution here as well, because we think that could be very informative also for selecting the right cancer interception drugs or taking those markers and putting them together into a panel to better capture the dysplasia. And there might be useful clinical information in there too, about how likely they are to progress to cancer, how aggressive their cancer will be, how invasive, those sorts of features, all things we're very interested in.

Host: Pretty cool. This is so fascinating. We don't have time to talk about the other aspect where you brought it back to cell by cell, like, where exactly these alterations are by doing this spatial transcriptomic piece, but a fascinating paper. For those of who are interested, look at the images and read. It's really admirable.

Well, thank you Sydney. This has been a very, very enlightening conversation and hopefully, some of the audience will be excited about this technology that may come to our bench one day soon. So, thank you very much.

Sydney Shaffer, MD: Thank you so much for the opportunity. It's been great talking with you today.

Outro: We would like to extend our sincere gratitude to today's guests for their valuable contributions and thoughtful perspectives. We also wish to thank our dedicated production team and the technical team at DoctorPodcasting. Special thanks to Jennifer Vepler for her dedication and skill in coordinating our episodes.

Please note the opinions expressed by our speakers are their own and do not necessarily represent the Perelman School of Medicine or the University of Pennsylvania. Thank you for listening, and we look forward to welcoming you back for future episodes.