Advances in translational cancer research  

Albert Koong, M.D., Ph.D., Chief Scientific Officer Hi, I'm Dr. Albert Koong, chief scientific officer at UT MD Anderson. I'm here today with Dr. Nick Navin, chair of Systems Biology at UT MD Anderson. And this is the Cancerwise podcast. Thanks for joining me today, Nick. 

Nick Navin, Ph.D., Chair, Systems Biology Thanks. Great to be here. 

Albert Koong, M.D., Ph.D., Chief Scientific Officer Today we're going to talk about advances in translational cancer research and more specifically, how emerging biomarkers can help inform and accelerate the pace of clinical trials, specifically to bring more therapies to our patients faster. A lot of this is biomarker driven, and one of the technologies that is utilized frequently is ctDNA. Maybe you can give us a little bit of background on that, and what you think is the emerging role. 

Nick Navin, Ph.D., Chair, Systems Biology CtDNA or circulating tumor DNA is DNA that's shed into your blood from the cancer cells. It's actually also shed by normal cells, but cancer cells and tumor, many of them are dying. They also secrete a lot of things, so they're able to get that DNA into your blood and into your circulation. And then an oncologist can take a blood sample, which is a very noninvasive procedure, and then they can look at that DNA and understand which mutations have occurred, and they can track those over time. But it wasn't until a lot of research and new technologies like PCR, which can really amplify DNA a lot, were we able to start to look at mutations from tumors in the blood. It wasn't until about 2016 where the first assays were developed for clinical treatment that was in EGFR mutation for lung cancer. And since then, there have been a whole bunch of companies that have developed both single mutation tracking assays and panels of genes that you can look at over time to understand how the tumor is changing dynamically. I think it's one of the great examples of where basic science studies of DNA in the blood and understanding of cancer evolution have led to a clinical assay and a way that we can really track things in patients over time to improve their therapies and therapeutic strategies. So, it's really quite a valuable assay that's being used a lot now in the clinic. 

Albert Koong, M.D., Ph.D., Chief Scientific Officer So, we're really talking about is this era of precision oncology and using a ctDNA as a way to determine how to give certain therapies, how long to give certain therapies, and the effectiveness of certain therapies. So, I think what we're really saying is instead of doing thousands of patient trials, perhaps we can get to the same answer, but with maybe 10 or 20 or 30 patients with deep sampling and multiple samplings throughout the course of treatment. So, as you think about the future of, of clinical trials and advancing oncologic care and new therapies, ctDNA is certainly going to be part of that. But are there other ways now that we can profile a tumor in the blood? And how are these assays being used? 

Nick Navin, Ph.D., Chair, Systems Biology Oh yeah. Absolutely. There are also now methods to look at things called exosomes or small vesicles, and they can also trap other types of molecules like RNA. So, that can be used to understand how patients are responding to therapy. And basically, what we're looking at is whether those mutations are going away in the blood or are new mutations arising over time. And so, that can lead to changes in how an oncologist might decide to treat that patient, which can be very valuable. The other thing is that in smaller trials, like Phase 1 trials where you might have 20 or 30 patients, we also collect core biopsy samples. And in the past that was a very small amount of tissue, and there wasn't that much you could do with it other than look at the pathology. But these days you can measure thousands and thousands of molecules from a very limited amount of material and even material that's kind of standard archival, what we call FFP materials, and so, that gives us a lot of information that we can collect from a patient over time. And so, even from a small number of patients by this kind of deep profiling, we often call this multi-omics because we're not measuring one thing. It's not just looking at DNA. We're looking at DNA and RNA and protein and also histopathology. We can really learn a lot about how patients are responding to new types of drugs. So, I think that's really a big advantage here at MD Anderson that we have so many of these trials and people on the research side that have these new technologies that can look at these samples and really learn a lot of information about who's going to benefit from which therapies. 

Albert Koong, M.D., Ph.D., Chief Scientific Officer So, these translationally focused trials that we're talking about that can more rapidly advance cancer therapy is really tapping into the quote unquote omics. What exactly does it mean when we talk about omics? 

Nick Navin, Ph.D., Chair, Systems Biology Omics is really just a set of technologies where we try to measure a lot of different things at once. So, instead of looking at a single biomarker or a single gene. Genomics is looking at DNA and across the whole genome. So, all the 20,000 genes that we have in the human genome. Transcriptomics is looking at all the RNA species. Proteomics is looking at all the proteins. There's also metabolomics and there's many types of omics. So, what it really is, is this kind of more unbiased profiling of all the genes or all the pathways and sampling those to learn about cancer as kind of a whole system, and then trying to integrate that data back together so we can understand, you know, how changes in DNA lead to changes in RNA and changes in protein, ultimately that affects the cancer cells and how they grow and what their therapeutic vulnerabilities are. So, that's what systems biology is all about. We try to look at the whole system rather than a single gene or a single pathway. I think the technologies are becoming more mature, more affordable, and probably the most important that they're compatible with archival clinical samples, which was a big barrier for a long time. 

Albert Koong, M.D., Ph.D., Chief Scientific Officer So, really, you're talking about perturbation of one part of the system, and what does that impact on the rest of the system? And although we're focusing on a research discussion and how that can drive clinical trials and clinical advancements, are there examples now where omics are being used routinely in clinical care and we have a proof test that any patient can get a certain profile and how that can predict clinical outcomes? 

Nick Navin, Ph.D., Chair, Systems Biology Yes, there are examples. I would say one that's used a lot are these cancer gene panels. So, at MD Anderson we had our own cancer gene panel that we developed. You're essentially looking for mostly point mutations there at the DNA level for hundreds of genes. And if you have certain mutations, for example, mutation in EGFR or mutation and KRAS or in BRAF, that means you could be eligible for a targeted therapy that is specific to that mutation. And so, by getting those profiles of your tumor, patients are able to make decisions about therapies that they might do in addition to their chemotherapy or radiotherapy that target specific mutations for their cancer type, and in some cancers, those have been shown to be very effective. 

Albert Koong, M.D., Ph.D., Chief Scientific Officer Yeah, that's hard to talk about this multi-dimensionality of data without bringing in a conversation related to AI. So, in your estimation, what would be the role of AI in helping clinicians integrate these very complex data sets that are changing dynamically throughout the course of therapy and ultimately leading to a readout that helps guide the clinician’s decision making and guide patient care kind of questions. So, what does that future look like? 

Nick Navin, Ph.D., Chair, Systems Biology Yeah, I think artificial intelligence is beneficial both on the clinic side and on the research side. On the research side, AI methods are very good at coding, for example. So, they just speed up our ability to analyze and process data. And on the clinic side, what's exciting I think is something that we call a agentic AI. You can think of this as when you use AI, you often just kind of ask a question. You get an answer. Agentic AI is you delegate a task or a job, and then the AI agent is going to go out there and work on it, sometimes for a long time, to complete that project. AI agents can be used in the clinic to potentially monitor a patient constantly. So, as new test results are coming in, as new data on mutations is coming in, there are changes in any metabolic levels or blood markers, they can immediately flag that and flag the oncologists. They're able to really, I would say integrate lots of different modalities. And that will hopefully in the future really inform how patients are treated and diagnosed and caught early when they start deteriorating. I think those are, those are some exciting applications. What's happening right now, I would say more, is AI has had a huge impact on pathology. There, AI is very good at looking for patterns in data that has largely been qualitative for a long time. Pathologists look at a slide of tissue from a biopsy, and then they try to classify the cancer type and perhaps what the treatment might be. AI is trained on thousands or tens of thousands of tissues, and they can classify features that the human eye can't even see. So, right now in pathology, some AI methods are being used to both augment pathology, which means they can annotate different areas that the pathologist should pay more attention to. And they're also very useful for things like risk stratification. So, being able to parry out which patients might be high risk or low risk. So, they're having an impact there. So, there's no doubt it's having a big impact I think across many areas of oncology and research. Of course, many other areas outside of the cancer field as well. 

Albert Koong, M.D., Ph.D., Chief Scientific Officer Yeah, you make some great points, Nick. I mean, for, for me, the future really is how can we harness artificial intelligence to integrate all of these levels of data that you're talking about? Genomics data, all the omics data, pathology data, imaging data, and even just routine clinical data, put it all into an algorithm that will help predict how patients will respond to certain therapies and, more importantly, which ones are most likely to work and which ones maybe would lead to certain toxicities that you would want to avoid. And that would be a super useful tool for our clinicians to utilize in order to decide what approach would be the best for a specific set of patients. And that's really the ultimate in personalized medicine and precision oncology that we hear a lot about these days in the oncology world. I think one of the applications is matching patients with the appropriate clinical trials. The inclusion and exclusion criteria for clinical trials is so complicated. Perhaps the most efficient way to do it now is to utilize AI. And they would highlight and flag patients that could potentially be eligible for clinical trials. Similarly, on the patient side, there's lots of searches now being done with AI. Patients can seek out clinical trials, perhaps those that are most likely to benefit from the particular therapy and matching that with the disease stage and disease type. So, I can see a world where AI kind of brings patients and clinicians together in a very efficient manner so that we can more efficiently run these trials and more quickly get to the outcomes, which is really to advance cancer therapies for, for our patients. 

Nick Navin, Ph.D., Chair, Systems Biology I mean, I think some other examples I know about in the clinic are things like just taking notes while meeting with patients and documenting those and then being able to search those later as one of the data modalities. I've also heard about it being used for things like endoscopy, where real-time AI imaging can find premalignant lesions and areas that the oncologist might want to biopsy so that they can look later at pathology and see if it really is a pre-cancer or not. I think there's many areas where AI is already impacting the clinic and many more to come, I think. 

Albert Koong, M.D., Ph.D., Chief Scientific Officer Yeah, AI is very good at highlighting things that clinicians should focus on. So, AI-assisted diagnosis, whether it's in pathology or in radiology or endoscopies as you mentioned, I think can only enhance the diagnostic accuracy. What are some of the most exciting technologies coming out of the lab? 

Nick Navin, Ph.D., Chair, Systems Biology I might be a little biased here, but I think methods that we work on a lot for single cell genomics are very exciting. We have methods to look at single cell DNA measurements. We can look at single cell RNA measurements, single cell protein measurements. And more and more now we can do multi-omics in single cells. So, from the same cell we can look at DNA and RNA, and that really helps us understand how the genetic alterations led to a phenotypic change in that cancer cell or a change in how it behaves. That's very powerful. I think spatial transcriptomics and spatial genomics in general is really accelerating fast. In the past, in pathology, we generally more qualitatively look at images of cells and tissues. And that's been very powerful and very useful in the clinic for for decades. But now from that same little tissue section that we cut from a biopsy or a surgical sample, we can measure tens of thousands of molecules, DNA molecules, RNA molecules. I think that's really going to transform pathology in the future. And that type of data is very ripe for deep learning methods, artificial intelligence, things like that. So, it's really quite impressive, I think the amount of information that we can collect. Also, I think we're some of the new technologies that are exciting too is looking at not just the DNA and RNA of a cell, but really its morphology and its morphometrics. And so, we have techniques now where we can do this very rapidly with things like microfluidics, nanowells, technologies like that. And so, being here at MD Anderson, what's exciting is we can not just develop these technologies, but we can try to bring them into the clinic and translate them and figure out where they can have the biggest impact on patients. That's very exciting. They have to be cost efficient. They have to be reproducible. But I think they really have a lot of power to improve things like diagnosis, help risk stratification, and discover new drug targets for treating things like metastatic disease. A lot of these technologies can be used on human tissues directly, and so that has the advantage of then being a much faster route to translate that information back into, into the clinic to help patients. Human tissues have been limited in the past. When you take a fine needle aspirate or a small core biopsy, which is just a little cylinder of tissue. In the past, it was hard to run a lot of these technologies, and that's really changed just in the last few years where we can do a lot with just 10,000 cells or 100,000 cells, which is just a little sliver of tissue. Working with human tissue samples and looking at them longitudinally, I think there's really a lot that we can learn there. And then we can always go back into the preclinical models, cell lines, things like that, and try to validate our results and explore the biology a lot more. But in terms of having a fast path to get information that we can use back in the clinic, I think working directly in human tissue samples is really a good route to go. 

Albert Koong, M.D., Ph.D., Chief Scientific Officer When you look at RNA, you're really looking at a dynamic profile of what's going on in the tumor, right? So, as you think about imaging or other kinds of modalities, a lot of those are more static. You're taking a single snapshot. But with RNA it's constantly changing. And you can get repeated biopsies that, in theory, should allow us to become even more precise in how we think about the tumor and the best way to treat patients. 

Nick Navin, Ph.D., Chair, Systems Biology Yes, absolutely. Longitudinal sampling can help with that a lot. Also, looking at single cells can help a lot because you know within one tissue every cell might be in a slightly different state. But you can kind of reconstruct those dynamics or those trajectories and understand we often refer to this as plasticity. It's kind of the ability of a cell to change its behavior or its phenotype quickly in response to things like therapy. So, yeah, I think we can learn a lot about that actually from single cell technologies as well, even from a single time point sample. But collecting dynamic information from these smaller trials, I think that's really exciting. And we've been starting to work a lot on samples from smaller trials and trying to do very deep profiling from longitudinal samples. 

Albert Koong, M.D., Ph.D., Chief Scientific Officer So, that's a good segue into spatial transcriptomics, trying to understand what's happening on a cellular basis and the interactions between individual cells within a tumor microenvironment or within a tumor biopsy that you might be able to see this. Can you describe that process and how we can use that data to guide clinical trials and clinical decision making? 

Nick Navin, Ph.D., Chair, Systems Biology So, spatial transcriptomics is a technology where we can take a tissue section. So, you can imagine you have a tumor, you cut through it and you're kind of looking at one section through that tumor. And within there, there's a lot of different cells. You have the cancer cells, but it's not just the cancer cells alone. There are immune cells that are surrounding it everywhere. The T cells. The B cells. The myeloid cells. There's the connective cells, so those are we call fibroblasts. And there's the cells that bring blood in. So, the vascular cells, the venous cells, arterial cells, the capillary cells. And the tumor cells are very good at reprograming that whole microenvironment. What that means is they're really talking to the cells and changing their normal functions. They want more blood. They want to create like this fibrosis, which is kind of hardening of the tissue. They want to get rid of most immune cells because the immune cells are out there to eliminate the cancer cells. So, there's this whole milieu of cells in the tissue, and understanding every little cell and how it relates to the cancer cells is really quite important. For example, with immunotherapy, you'd like to know if T cells, for the most part, are able to get into the cancer cells and access them where they make direct contact to kill those cells or not. And in many cases, the cancer cells are very good at putting up barriers, using other cells to exclude T cells. And so, if you look at spatial transcriptomics and you see that the majority of T cells are being excluded, that might mean that the immunotherapies might not be as effective as well. So, having that spatial information is, is really important. In the past these assays have been kind of expensive. And so, it's been hard to look at, you know, more than 10 samples or 20 samples at a time. But the technologies are rapidly evolving. Lots of commercial companies that are developing these technologies, and they're changing every six months, but that means the prices are coming down. And before we could only measure a few hundred genes, and now we can measure the whole transcriptome just from a single tissue section. So, I think it's a really exciting area of new technologies. Here at MD Anderson, we have a lot of investigators that are working on this. We hope that these will be used in the clinic as well to really improve the diagnosis and treatment of cancer patients. 

Albert Koong, M.D., Ph.D., Chief Scientific Officer Yeah. That's amazing, Nick. And just to think, this technology has only become available, I would say, in the last few years, really. And many of our clinical trials now is incorporating this spatial transcriptomic data into the trial itself to try to understand, for example, patients getting a checkpoint inhibitor and you have a pre- and a post-biopsy. And if you see or don't see an immune infiltrate into the tumor, that might guide your decision making and perhaps indicate how effective treatment is going to be. Even before we get to the endpoint of, let's say, overall survival or response rate, we would know much earlier in the course of the therapy if this treatment is working, using technologies such as this spatial transcriptomics that you just described. And I think in the coming years, this type of data will become more and more available, more generalizable, and will accelerate our cancer therapies even faster than what we're seeing today. 

Nick Navin, Ph.D., Chair, Systems Biology Yes, absolutely. The future is bright. 

Albert Koong, M.D., Ph.D., Chief Scientific Officer Thank you, Nick, for joining me today. It was a great conversation. 

Nick Navin, Ph.D., Chair, Systems Biology Thanks for having me. I had a lot of fun. 

Albert Koong, M.D., Ph.D., Chief Scientific Officer And thanks for tuning in today. If you enjoyed this episode, be sure to follow or subscribe on Apple Podcasts, Spotify, YouTube, or wherever you get your podcasts. And don't forget to comment or review. For more information or to request an appointment, call 1-877-632-6789 or visit MD Anderson.org. Thanks for listening to the Cancerwise podcast from UT MD Anderson.