All right, we're going to go ahead and get started to keep us on time. So thank you everyone for staying to the last day in the afternoon of this conference. So I'm Jeffrey Thompson, I'm Assistant Professor of Medicine at the University of Pennsylvania. Today we'll be talking about what the pulmonologists have collected as far as lung cancer biomarkers. We're fortunate to start off with Dr. Otis Rickman, Dr. Rickman's Associate Professor of Medicine at Vanderbilt, and he's an interventional pulmonologist and is the Director of the Interventional Pulmonary Program at Vanderbilt, and he's going to lead us off with talking about lung cancer mutations, optimizing the biopsies. Thank you, Dr. Thompson. And then thank you all for your sacrifice. I know how much pulmonologists love coming to molecular marker talk at the end of the meeting after lunch, and so we're going to leave the lights up really, really bright. And if I do see anybody nodding off, I'm going to call you out. But anyway, so my name is Otis Rickman, and thank you, Dr. Thompson, for that introduction. And we're going to make this as exciting as we can. And the other thing is that my name today is not just Otis Rickman, it's Otis Nicole Tanner Rickman, because I got the baton passed to me for this talk a little later than usual. So we're going to do the best we can. So these are my disclosures. Unfortunately, I don't have any big AstraZeneca, Sanofi, any of those big lucrative consulting contracts, but I am open. I do have a lot of device contracts, but nothing really that pertains to this, and it should just point out that this session is supported by Merck, Sharpen Dome, Sanofi, Sanko, others, which obviously have an interest in this. What we're going to talk about today is, we've all heard it, we've all been on our tumor boards, and there's really many opportunities to mess this up, and to delay care for our patients with lung cancer. And we're going to talk about our role as pulmonologists, as interventionalists, as thoracic surgeons, who are sometimes the first patient, the first person to touch these patients with a new diagnosis of lung cancer. And then the last thing that you want to hear as a proceduralist is, you didn't get enough tissue, right? You know, you hear that, and it's like, oh man, it just like hurts my pride. And so you want to make sure you get enough. I'm going to tell you how to do that today. So as pulmonologists, why do we need to know this? And people are going to talk about this, you've probably heard it at all meetings, that basically the times have changed. And now numerous gene alterations have been identified that impact therapy. And also, we have things, we know that the different types of therapy that you want to avoid whenever you know what the molecular profile, or what the histology of a patient is. These are all the things, I'm not going to talk about them, just to say that there's like eight or nine drugable targets now. We have a bazillion drugs that are out there now, and of course we have the molecular targets as well as the immunotherapy targets for, really for locally advanced and metastatic disease. These are the NCCN guidelines that, and notice on here on these NCCN guidelines, I'll come back to this several times, this is for advanced or metastatic disease, okay? And we'll kind of explore a little bit this sort of what our role is in local disease. The thing that's important that you see here is that molecular testing is recommended for all adenocarcinomas, okay? And that molecular testing shouldn't be a targeted panel anymore, it should be these next-gen sequencing where they're looking at 50, 100, 1,000 genes. That's what's recommended by the NCCN guidelines, which are the most common guidelines used for treatment of lung cancer in North America. And then for squamous cell, it says consider doing the treatment, but you don't necessarily have to do it for squamous. I'll say that in our institution, and one of our speakers today, Dr. Lovely, can really expand on that a little bit more. She's one of my oncology colleagues. So here are our societies, and they all basically support practices that promote biomarker testing. And so the ATS, CHESS, and then the American College of Radiology. And so the interventional radiologists, the pulmonologists, are going to be the ones that are getting these pieces of tissue for folks most often. There are multiple methods for acquiring these samples, and what we're going to focus on today is bronchoscopy and what its role is. And so, right, if you're a carpenter, everything looks like a nail, so we're all carpenters in here. And so the nail is our bronchoscope, and so that's what tool we're going to use to employ I will say that we also have a little bit of a role here for fluid cytology, and don't underestimate the power of the pleural effusion for getting tissue on these. And so if you have your lab set up and you have your processing set up, you can really get a nice specimen for next-gen sequencing off a pleural sample. In fact, Dr. Lovely, one of her cell lines that she's made immortal was off of a pleural effusion that I helped her with whenever she was a research fellow. So anyway, off we go here on tissue acquisition and management. And so tumor testing has been primarily focused on FFPE, formalin-fixed paraffin-embedded tissue. And it should be noted that testing on these cell blocks is not included in the FDA approval for some of these tests, but it probably will be. And so that's what you're going to see in the guidelines, is that it's not recommended. But I'm going to show you some data that suggests that it's perfectly adequate, these specimens that we give them. And then also in the NCCN guidelines, they point out that it's a major limitation of obtained tissue molecular testing with minimally invasive techniques. And I put this slide at the end because I didn't know if I was going to talk about it, but I'm going to talk about it now anyway, is that these guidelines, so the NCCN guidelines, how many pulmonologists are on the NCCN guidelines? Anybody know? So up until three months ago, there was zero, okay? How many interventional radiologists are on the NCCN guideline panels for non-small cell? One, okay? And the people that obtain 90% of these specimens don't have a seat at the table for writing the guidelines. And so we were able to get one of our colleagues, Dr. Maldonado, on the panel about a couple of months ago. And so hopefully we're going to get a little bit, you know, what's that Rodney Dangerfield again? No respect! So maybe we'll get a little bit more respect here. So here we go. This is important. So you don't want to, you know, these are expensive, right? And so you don't want to use some, you know, fly-by-night lab where you found on Amazon or something like that and send your specimen off. You want a CLIA-certified lab that really does know how to handle your specimens. And you also want to know how many genes they do, how much tissue that they do. We'll get into that a little bit. And so the main thing is that your tumor sample quantity has to be sufficient to support multiple diagnostic analyses. What are the minimum diagnostic analyses we need to do? Well, we need to look at it morphologically under H&E. Can we tell? Is it a non-small cell? Is it a small cell? Can I tell just by looking at it? Is it an adenocarcinoma? Is it a squamous cell carcinoma? And then I should judiciously use immunohistochemistry to protect this small specimen. And so if I know that this is a, you know, a 65-year-old smoker with a speculated two-centimeter left upper lung nodule and no history of other cancer, and my pathologist goes through and he stains it for synaptophysin and some kind of colon marker and a prostate marker, and then throws a breast cancer one in there just because he's using up all that specimen. He or she's using up all that specimen. And so you really got to talk with your pathologist for doing it. They can classify 90 percent of these cancers with two stains. So with a TTF1 for adenocarcinoma and a P50, P60, P53, I think it's something like that anyway, for squamous cell carcinoma. And then sometimes they'll throw in a napsin A or something like that on top of it. But anyway, so just a couple of stains on the immunohistochemistry and you can really go a long way. All right. So some challenges with tissue acquisition. A lot of times we get comorbidities, poor performance status. They're on some sort of anticoagulation that can't be stopped. Patients don't want to get a biopsy. There could be tumor factors, so it's a completely necrotic tumor. You can't get live tissue. It's not accessible for a safe biopsy. And it's very interesting in some of these about patients that are unable or refuse to undergo a biopsy. So 3 to 25 percent can't undergo biopsy for health reasons, and about another 2 percent refuse it just because they don't want to. And so this has led to a lot of, if you can get them to get the biopsy, make sure you do a good job and get enough tissue. But if you can't, there are other stuff that will be talked about later in this talk with liquid biopsy that certainly is getting a lot of attention right now. Yeah, we talked about that already. All right. So this was going to be an audience response, but just somebody to date. Of these types of tissue acquisition methods, which one has been able to provide the most adequate material? Is it EBUS-TBNA? Is it bronchoscopic lung biopsy? Is it a CT-guided FNA or core biopsy? Or is it a bone biopsy? What do you think? Somebody shouted out something. So bronchoscopic lung biopsy. So going out, we arguably include EBUS-TBNA in that. So EBUS-TBNA has been used for molecular analysis, and now we're at the 10-year anniversary of some of these original papers that were published back in 2011 and 2012. And so again, these weren't next-gen sequencing. They were sort of targeted papers for EGFR and BRAF out and some other ones. But 72% to 99% adequacy in those, or 80% of cases. And so it's been there for a long time. So we know that we can get a lot of tissue. And you're going to hear, if you go to your tumor board, what does your oncologist want? Hey, can you get me a core, right? Okay. All right. And well, you can tell them, no, you don't need a core. I can get you something better than a core, is I can get you a cell block that's going to be adequate for testing 99% of the time. And so here's the EBUS for genetic testing, initially evaluated on single genes, pulled analysis, 28 studies, 95% of the time was sufficient. Another 12 studies, 607 patients, ALK, 95% of the time it was adequate. And then smaller studies for ROS1, 83% of the time, just EBUS-TBNA. And then what about next-gen sequencing? So this is a big meta-analysis that was done. And basically, if you skip all that other crap there and go down there to the bottom, it says the modeled yields were 77%, 86%, 91%, and 94% at 3, 4, 5, and 6 passes. So I know somebody was going to ask me this in a question, so I already answered it. Six passes is how many that you needed to do. All right. So the DNA requirements, it depends on the NGS platform you use. We talked about that. So do you send it to Tempest? Do you send it to Foundation One? Do you send it to BRLI and probably some other ones I can't remember, but anyway. And large panels of more than 200 genes can require up to 50 nanograms of DNA. The smaller panels of 50 genes, like Oncomune and some other ones like that, they may require 1 to 4 nanograms of DNA. So anyway, so the more genes that you try to test for, then the more tissue is going to be needed. So we talked about this already, about the immunohistochemistry, about the P63 and P40 for adeno and the TTF1, sorry, adeno is TTF1, P63, P40 is for squane. So preserve that tissue. This is interesting that eBus may be better than bronchoscopic biopsy and possibly TTNA for molecular markers. And it shows that eBus TBNA has the highest, right here, has the highest yield for multiple biomarker testing compared to bronchoscopy or even CT-guided core biopsy. And then if we look at this eBus TBNA versus CT-guided biopsy versus bone biopsy, if you stay in this, here's the lung and then here's the lung image-guided core biopsy. If you scroll right down to the bottom, you can see that basically eBus TBNA outperforms all the other modalities, even core needle biopsy. And you really got to be careful if you're sending somebody for a bone biopsy because once you decalcify it, you destroy it. And they're almost never able to recover from that. So those highest failure rates are with CT-guided core biopsy and bone specimens. So eBus for next-gen sequencing, these 340 of 469 genes, 19 were clinically validated. And then this is interesting, the success of doing this. So this was at, I think this was the Memorial Sloan-Kettering Impact Panel. And whenever they first started doing it, they were getting an adequacy in 76%. And then the final, the third cohort, it was up to 92%. So they do get some benefit as you learn how to do this in your institution over time and process these specimens better. This is a slide from one of my colleagues at the University of Chicago, Dr. Tim Ragoo. And they can actually, doing this OncoScreen and OncoPlus, they can go in and do a laser dissection on the actual smear, the smear that was done. And so they just cut out the tumor, and they're able to get 85% on smears. Notably, it destroys the slide. So that slide's gone. You can't use it anymore. But just on one slide, they can get enough tissue to be able to do this testing. So what about needle gauge? That was the other question. So six-pass is right. And you're going to ask me also about what needle should I use, OK? Doesn't matter. Is the Cliff Notes version of that 22 or 25-gauge needle and mostly 22 or 21-gauge needle? Don't ask me about the 19-gauge needle, because I don't know. But if you extrapolate from the E-Bus literature, the 19-gauge needle didn't make a difference. Either it got you a bigger specimen, but it was mostly blood. And then whenever you do the needle, the other thing that's frequently asked, so we do six-passes, and then you need to do 10 to 15 strokes on that pass to get it. And I can't tell you for next-gen sequencing whether you should use suction, whether you should use the stylet or not. I can tell you for diagnostic yield, it doesn't matter. You can use whatever you want. So probably going to be the same for NGS. Peripheral nodules, OK? So stage 1, stage 2 lung cancers. The NCCN guidelines, just this most recent version 5, were updated because of the Checkmate 816, which is this trial here, to add in neoadjuvant nevolumib. So immunotherapy, not molecular targeted therapy, but immunotherapy before resection. So interestingly, in this trial, they didn't look at how much what the expression was for the PD-L1. They just randomly assigned people. But you could probably make the argument that if you were to do PD-L1 testing in the people who overexpress, it would probably be better. A little over here. The success rate for each of the bronchoscopic devices, I just want to point out here, we look at EBUS guide sheath, endobronchial biopsy, or EBUS TBNA. So we look here. So these are all peripheral nodules. And so you can see that you want to use the larger EBUS guide sheath. That means you're using a bigger instrument. And your DNA and RNA are 95-plus percent sufficient. The type of tool, this was one that looked at a cryoprobe versus a guide sheath. The cryoprobe seemed to be a little bit better. This looks at forceps, so a standard forceps versus a standard forceps plus a small forceps. And then just the small forceps alone. And so you can see here, 85-percent adequate, 82-percent adequate, and 70-percent adequate. And again, these were for peripheral nodules. You can do it for PD-L1 testing. It's great. And let's see here. We're going to move on. And then ROSE, should you use ROSE or not? ROSE does improve the adequacy for molecular testing based on this trial. Again, I don't know for NGS, but that was for a targeted gene panel. Immediate fixation is important. So you don't want to do these on Friday night and then let them sit in something all weekend. So you want to fix these immediately after you take them out of the patient. Delayed fixation results in loss of receptors. You want to fix them in a buffered formalin solution. Ethanol is okay for cytology, but is not good for histology. So put those specimens in formalin, and the formalin fixation should happen within six hours. But even within an hour is even better. So eBus TBNA, these are the take-home points, is good. Use whatever size needle that you want. Use ROSE and get enough tissue. And then finally, lots of guidelines of support to the use of cytology specimens for NGS in molecular testing. And don't over-process the material. And remember, so the final thing is bronchoscopy is king in this, because we can get a diagnosis, we can stage, and we can sort of set the platform and the path going forward for that patient. And we also set the timeline. So basically, you want to be a key member of your tumor board and of your lung cancer team. And so that you're listening to what the people are going to need. You're getting the tissue that the people are going to need to make treatment decisions down the road. So do a good job. Thank you. Great talk. I agree that molecular testing on bronchoscopy specimens is becoming increasingly important, particularly as molecular targeted therapy and immunotherapy is improved in the adjuvant and neoadjuvant setting. So do you do six passes on all your staging eBus exams for each node, or just one where you were trying to get material from that node? Great. So to sort of break that down then a little bit, so the way that we do all of our eBus with rows, and so if you were to do six passes on every node that you did, and so this is not the point of this talk, but you should be sampling every node that's above five millimeters in size, is that we will do two for rows immediate in the room, and then we'll put one in cell block, and then we move on. And then if we get a call of cancer, then we'll go back to that node and get our additional up to six passes, and put that into whatever specimen we're using that for. We actually put our specimens into RPMI, so the Roswell Park medium, and then the cells are kept alive, they're transported back, and the reason for that is if they wanted to do flow on those, that they still could, because if I drop them in satellite immediately or into formalin immediately, then flow cytometry is out the window. All right. Okay, thank you. All right, so I'm going to be talking about Beyond PD-L1 Expression, New Immune Markers for Lung Cancer. As I said, I'm Jeffrey Thompson. I'm an assistant professor of medicine in the section of interventional pulmonology at the University of Pennsylvania. Here are a few of my disclosures. So today, I'm going to talk about some of the limitations of PD-L1 as a biomarker in general. We'll discuss why some patients may not respond to immunotherapy. We'll review some of the emerging biomarkers of response to immunotherapy, and then I'll close out with some future directions and opportunities. So I probably don't need to tell anyone in this room that the treatment landscape for non-small cell lung cancer has evolved quite dramatically in the last decade. And this is really through two central pillars, one of which being molecular-targeted therapy, and the other, what I'm going to focus mostly on, is immunotherapy. An understanding of tumor-immune interactions and how tumor cells evade the immune system by upregulating these inhibitory molecules, and then we can block these immune checkpoints and enable cytotoxic T-cell killing of the tumors. This is just a graph showing the FDA approvals of drugs over the last 10 years or so, and you can see the yellow boxes of all the immunotherapeutic strategies that have been approved for lung cancer. And this has really changed the treatment paradigm for our patients. However, we know that still only a subgroup of patients respond to these therapies. We know that immunotherapy is associated with increased toxicity and increased costs. We're all very familiar with immune-related adverse events, particularly pneumonitis that gets referred to our office. And there's been described the hyper-progressive phenotype. So patients get immunotherapy, they have a rapid clinical deterioration. And so really, a key interest in the field has been to, how do we better identify these patients that are most likely to respond to these therapies, so that we can avoid unnecessary toxicities, reduce costs, and enable more appropriate therapies to be delivered? And currently, the most commonly utilized biomarker of response to immunotherapy has been tumor cell PD-L1 expression, and I've just highlighted some examples of PD-L1 expression here, with on the left, a patient with very low PD-L1 expression, and on the right, greater than 50% with a patient with very high levels of PD-L1 expression. Those with higher levels of PD-L1 expression would be those most likely to respond to immunotherapy. But PD-L1 has a number of limitations. If you see on the right, that even when we enrich patients for PD-L1 expression, still 40% to 50% of those patients don't respond to an immunotherapeutic strategy. In addition, there have been a number of clinical trials showing different immune-targeted agents have shown improved outcomes in patients independent of PD-L1 expression. And we know the various limitations of PD-L1 as a biomarker, and that this is a dynamic biomarker and exhibits both spatial and temporal heterogeneity. So this is a surgically resected specimen, and depending on where you biopsied this lesion, you might think that the patient might have anywhere from very low to high levels of PD-L1 expression, and it has profound treatment implications for our patients. Not to mention that the field has really been plagued by the fact that there are multiple PD-L1 assays out there, which has really kind of perplexed the medical community, causing a lot of confusion with, you know, different antibodies and different cutoffs used in different clinical trials. And fortunately, there have been a lot of harmonization efforts over the last few years trying to establish whether there's concordance across these various antibodies. And you know, the good news is, for the vast majority of these antibodies, you know, there is good concordance in PD-L1 expression, but there are some outliers, and it's important to understand, you know, what particular PD-L1 assay you use at your own institution. And I just want to highlight that one of the reasons that PD-L1 is a limited biomarker, it doesn't really fully capture the complexity of generating an effective anti-immune response. So this is really a complex, multi-step process that really involves, you know, the tumor presenting antigen to T-cells, those T-cells getting activated, trafficking into the tumor microenvironment, recognizing the tumor, and then inducing cytotoxic T-cell killing. And so this is a complex process, and I just want to spend the rest of the talk kind of highlighting some of the key elements that are associated with generating an effective anti-tumor immune response. And that is looking at tumor mutation burden. Can tumors, you know, generate sufficient neoantigens to be recognized by the tumor? Antigen presentation. Do they present antigen on their cell surface to be recognized by T-cells? And then looking at the tumor microenvironment, and is the tumor microenvironment conducive to activated T-cells? So I'm going to first start with tumor mutation burden. So the theory behind tumor mutation burden is that throughout our lives, through aging, smoking, you know, UV light exposure, that we acquired different genetic mutations in our genome, and then those mutations get translated into proteins that are recognized as foreign by our immune system and our neoantigens. And the theory is that the more mutations you have, the more neoantigens you will have to be recognized by the immune system, and that those patients would have a greater response to immunotherapy. And if you look at this bottom right graph, and this was a study published in Nature back in 2013, looking at the tumor mutation burden across a whole host of different tumor types. And you can see at the far right, far right of this graph, that lung cancer and melanoma are really at the top tier of those tumor types that have high tumor mutation burden. And interestingly, you know, those tumors tend to be very responsive to immunotherapy. And so can tumor mutation burden be a good biomarker of response to an immunotherapeutic strategy? So there's been a lot of interest in this. Several studies have looked at whether TMB correlates with PD-L1 expression. And you can see in the Venn diagrams here, looking at both tumor mutation burden as established by the tumor. And you can also look at tumor mutation burden by analyzing circulating tumor DNA. And circulating tumor DNA is a topic that Dr. Lovely will talk about here in a few minutes. But suffice it to say that you don't see much overlap between high TMB and high PD-L1 expression in patients. And so there's a potential that these could be complementary biomarkers. In this bar chart here, this was a study performed at Memorial Sloan-Kettering looking at a cohort of non-small cell lung cancer patients, demonstrating that those patients that have high TMB and high PD-L1 expression were the most likely to respond to an immune checkpoint inhibitor. And those patients with low TMB and low PD-L1 expression were the least likely to respond to immune checkpoint inhibitor, suggesting that these are really complementary biomarkers. Now this is a busy table, but this is just highlighting that a number of clinical trials have been performed evaluating whether TMB can predict response to immunotherapy in patients with non-small cell lung cancer. And the results have really been conflicting, with some studies showing that it does predict response, while others showing that it really doesn't. And then a major challenge that has really played the field for this to utilize this as a biomarker is that various different studies have used different platforms for TMB estimation, different cutoffs for what defines high TMB versus low TMB. And so it's been a real challenge to implement this clinically. And I like this editorial that was published in General Oncology a few years ago, just highlighting some of the pitfalls of using tumor mutation burden. And the major pitfalls I've mentioned already is that its prognostic and predictive ability has really not been prospectively validated quite yet. And that we need to identify reproducible cutoffs across studies before we implement this in clinical practice. And so I think tumor mutation burden is likely useful, but we need more prospective data and harmonization of cutoffs. So next I want to talk about antigen presentation. So a critical component to generating an effective anti-tumor immune response is the tumor has to present antigen on the tumor cell surface to the T cells via its MHC molecules. And I like this cartoon schematic here, showing the tumor in green, which its intact MHC class 1 molecules. And then over time, through selective pressure under the immune system, ultimately it down regulates its MHC class 1 molecules, thus evading immune surveillance and becoming unrecognizable to the T cells. And so it doesn't just do this in a cartoon. When you look at non-small cell lung cancer cases, non-small cell lung cancer frequently down regulates its HLA markers. You can see in the pie chart with the black, blue, red, and green pies there, how frequently non-small cell lung cancer cases down regulate HLA markers, thus evading the immune system. And this is estimated to occur in upwards of 30 to 50% of cases. And when we look at gene expression profiles that represent a tumor's ability to present antigen on the tumor cell surface, we see that loss of antigen processing machinery is associated with lower response to immunotherapy and worse overall survival. Very similarly, when you look at small cell lung cancer, small cell lung cancer classically down regulates its MHC class 1 molecules, thus evading immune surveillance. But when we focus on a specific subset of patients that maintain high MHC class 1 expression, and look how did those patients respond to immunotherapy, you can see that these patients with intact MHC class 1 expression and maintain the ability to present antigen on the tumor cell surface are those patients most likely to respond to an immunotherapeutic strategy. So I think antigen presentation is a really evolving potential biomarker of response, but further needs some more prospective data as well. And then lastly, I want to talk about the tumor microenvironment. So there's a number of different studies evaluating, you know, does a patient have a conducive microenvironment to responding to an immunotherapeutic strategy? And they have kind of interesting names for first of defining a cold tumor microenvironment. And a cold tumor microenvironment is one that really doesn't have any influx of CD8 T-cells, and you wouldn't think that that patient would respond very well to immunotherapy. Whereas a hot tumor microenvironment is one in which has a large influx of CD8 T-cells, a lot of chemokines, you know, enabling T-cell activation, and you would think that those patients would have the greatest ability to respond to immunotherapy. And so one of the ways where you can define whether someone has a hot versus cold tumor microenvironment is you can take that FFPE tissue and perform RNA sequencing and look for specific genetic transcripts associated with T-cell activation. And so when you do that, when you establish gene signatures associated with an inflamed tumor microenvironment, you can identify that, you know, patients with an inflamed tumor microenvironment identified by these gene expression profiles do indeed have improved response to checkpoint inhibitors. And this was a study published in Science a few years ago just highlighting that an inflamed gene signature is associated with improved overall response to immune checkpoint blockade across a variety of different tumor types. So some of the advantage of using these gene signatures is you're able to interrogate multiple cellular programs within the tumor. However, it has several disadvantages, that it already requires, you know, very limited amount of tissue. Different trials have used, you know, different gene signatures with different cutoffs, so there's no real standardization. And then the tumor microenvironment can be heterogeneous in a very similar fashion that, you know, PD-L1 expression is heterogeneous within the tumor microenvironment, thus limiting the clinical application of these gene signatures. And then lastly, I want to define a few what are called immune phenotypes. And so you can see in the top left here that an immune inflamed phenotype is a phenotype in which you have a heavy CD8 infiltration into the tumor. You would think that that patient would have, you know, reasonable response to immunotherapy. Whereas if you had immune excluded phenotype, which is the T-cells kind of sit on the perimeter of the tumor, or an immune desert when you look under the microscope and there aren't any T-cells to be found, that those patients would be least likely to respond to an immune checkpoint inhibitor. And so can we just stain for CD8 T-cells with the tumor microenvironment to predict response? And you know, this is a representative example where the CD8 T-cells are stained in brown here. And you can see that this is a patient who has, you know, a high amount of CD8 T-cell infiltration in their tumor microenvironment, where the patient on the right has virtually no CD8 T-cells in their tumor. And then when we look at these patients, the ones with high CD8 infiltration are the patients most likely to respond to immune checkpoint blockade. However, you can imagine that this is quite laborious, you know, having the pathologist, you know, individually count whether patients have high or low CD8s within their tumor microenvironment. And there's a lot of inter-observer heterogeneity within this. And so is there a better way, you know, to go about doing this? And I'll talk about some of the evolving, you know, technologies, which are changing quite rapidly. So this was a paper just published in Journal of Clinical Oncology just a few months ago looking at, you know, leveraging artificial intelligence to look at the slides and categorize the various phenotypes using machine learning. And so what they did was they took a large cohort of non-sponsored lung cancer patients receiving immune checkpoint blockade. Using digital pathology, they uploaded their H&E slides. And using a machine learning approach, categorized the patients into an inflamed, immune-excluded, and immune-desert phenotype, and then looked at their treatment outcomes. And so when you look at their progression-free and overall survival in response to immune checkpoint blockade, you can see in the red here that the patients with the inflamed phenotype had significantly improved progression-free and overall survival compared to patients with immune-excluded or immune-desert. So kind of a promising technology of kind of the next wave of leveraging, you know, technology and machine learning to improve, you know, biomarkers of response. And then lastly is enhanced digital imaging is really changing how we evaluate these patients. And it's really enabling a more comprehensive characterization of the tumor microenvironment. So not only we're just looking at CD8 T-cells within tumor microenvironment, but with, you know, more complex, multiplex imaging technologies and digital pathology, we can interrogate the various immune cell subsets within the tumor microenvironment and look at cell-cell interactions to understand further mechanisms of response and resistance. And I think it's these technologies that are really going to facilitate, you know, developing optimized immune strategies for our patients. So a few things that I'll leave you with that, you know, one of the ones that Dr. Rickman mentioned is that as pulmonologists, you know, why do we, you know, need to know this? Well, I think it's just important to highlight that the amount of, you know, biomarkers are rapidly evolving and we really need adequate tissue in order to do these studies. And multiple studies have shown that in about 15 to 30 percent of lung cancer cases that the patients don't have sufficient tissue for routine biomarker testing. And biomarker testing is only becoming increasingly important not just in the advanced stage setting, but in the early stage setting as well. And here on the right, I don't think it'll be a single biomarker that's going to win out to supplant PD-L1 and be the ultimate biomarker to predict immunotherapeutic response. But I think moving forward, it'll likely be a combination of various biomarkers using, you know, tumor tissue and blood analytes and combining, you know, various biomarkers in order to develop really optimized therapeutic strategies for our patients. So I'll just leave you with, you know, the use of checkpoint inhibitors has really transformed the treatment landscapes for our patients. PD-L1 expression by immunohistochemistry is the most commonly used biomarker response, but has a number of limitations, which I pointed out. And the development of biomarkers response to the checkpoint blockade is a rapidly evolving field but needs further standardization in prospective trials. But I think it's likely that a combination approach will be essential for achieving precision medicine for our patients. And that pulmonologists should really ensure that we have sufficient tissue and blood for such biomarker analysis so that each patient can be really considered for precision medicine. With that, thank you. Thank you. So, we'll invite Dr. Lovely here, who is, we're very fortunate to have a medical oncologist here. So, Dr. Lovely is an associate professor of medicine at Vanderbilt University and is the co-leader of the Translational and Interventional Oncology Program at Vanderbilt University, so thank you. I'll get you a seat. And she's super awesome. But absolutely not as awesome as Otis and all my interventional pulmonologists, so I'm giving away my final slide before I even start. So hi, everyone, it's such a pleasure to be here today. I have to say, in 16 years of being in Nashville, I've never been to the Music City Center, so thank you so much for having me and for letting me be part of this conference. So I am a medical oncologist, and I specialize in treating lung cancer, and I'm also a physician scientist, and it is really my honor to be here today and to work here at Vanderbilt and in this great city, and I hope that you all have an opportunity to enjoy Nashville while you're here. So I think I got one of the most provocative titles for any session in this entire conference, Cell-Free DNA, Will Liquid Biopsy Replace Pulmonologists? So I'm gonna start with the end, absolutely not, but we love liquid biopsy in medical oncology, and I hope to show you why. So these are my disclosures since 2012. So the objectives of this talk are really to review molecular alterations in lung cancer. We, in 2022, absolutely cannot make treatment decisions for our patients without understanding on a very granular level biomarkers in their tumor tissue, to describe use of plasma-based liquid biopsy in current clinical care, to analyze advantages and disadvantages to liquid biopsy. It's wonderful, but it's not perfect, and to describe future uses of liquid biopsy in our cancer care continuum. So in 2022, in molecular alterations in lung cancer, and I have to say, this is really the heart of my entire research program and the heart of all the patients that I treat in clinic, we currently have 10 FDA-approved genomic-driven biomarkers in addition to PD-L1 expression that we use to make treatment decisions for our patients. These are largely, but not exclusively, found in lung adenocarcinoma, and if you look at this pie, this is a representation of the frequency of some of those markers, and you've probably heard us talk about these at molecular tumor boards or just regular thoracic tumor board. These mutations are all tied to specific, often oral, therapies that we give our patients. They occur at varying frequencies, with KRAS mutations, for example, being in about 20 to 25% of all lung adenocarcinoma, and things like ROS1 fusions being in 1% or less. However, regardless of the frequency of these alterations, it is critical that we actually test for them in our patients because they allow us to select four different therapeutic options, and so when we think about our clinical toolbox in medical oncology, we have traditional cytotoxic chemotherapy, we have immunotherapy, which you just heard about, a huge and very quickly growing field, and we also have all of this enormous field of genomic-driven targeted therapies, which require us to know in advance how do we select the right patient for the right therapy based on the specific properties of their tumor DNA. So, again, the identification of tumor-specific genomic variants is absolutely central into guiding our treatment decisions across all stages of disease. This is not just for patients with metastatic lung cancer anymore. This is for early-stage disease now, too. So how do we actually identify these genomic biomarkers? Well, we have a couple different ways we can do that, and in the traditional way, it would be tumor biopsy, where the tumor tissue is sampled. All of your expertise, you know far more about obtaining tumor tissue biopsies than I do, but right now, in 2022, we have the ability to also do liquid biopsies, which is gonna be the focus of this discussion. Both of these can be used to extract genomic materials, such as DNA and RNA, from the tumor, and we typically use this genomic material to do a multi-gene panel next-generation sequencing assay, and there are multiple different assays that we can use. They can either be done on an institutional basis or for commercial send-out vendors, such as Foundation Medicine or Garden Health or Tempest. There are a variety of different vendors that do this. If you're interested in learning more about these panels, please find me afterwards. I'd love to chat with you about them. These are typically hundreds of genes with full exonic sequencing and often intronic sequencing as well, so it's a huge amount of information that we're getting every time we send one of these biopsy NGS studies. So what is a liquid biopsy exactly? So normal tissue turnover results in the release of DNA into the bloodstream. This DNA is called cell-free DNA or CF DNA. The component of cell-free DNA in the plasma that is from the tumor is called circulating tumor DNA or CT DNA. People often use those terms interchangeably, but they're not. CF DNA, cell-free DNA, has both tumor DNA and normal tissue DNA. Circulating tumor DNA is referred specifically to that DNA that's being released from turnover of the tumor. There are many different analytes that we can actually find in blood, not just CT DNA, such as circulating tumor cells, extracellular vesicles, and cell-free RNA. For the purposes of this discussion, we're largely gonna focus on CT DNA because that is what we use as standard of care in our clinical practice right now. And in the current state of the art, we can use this cell-free CT DNA to look for different variants, different DNA variants or different RNA variants that we can find that are associated with the tumor, such as single nucleotide variants, insertions, deletions, chromosomal rearrangements, and even epigenetic changes. So medical oncologists, I will have to say, the uptake of liquid biopsy has been quite rapid. And why is that the case? Well, liquid biopsy actually addresses some of the issues that are of concern when we think about tissue biopsy. So again, in no way, shape, or form do I think that liquid biopsy, bless you, will ever replace tissue biopsy. But the things we love about liquid biopsy in our clinic, first, it's a minimally invasive blood draw that can be done in a medical oncology clinic with a rapid turnaround time. So if I see my patient today in clinic and I send for a liquid biopsy, in about seven to 10 days, I'll have a result. That is absolutely so helpful when you're thinking about treating your patient, because instead of waiting potentially three weeks to four weeks to get the tissue biopsy, have that block sent to the external vendor, have the DNA extracted, the DNA sequenced, and then sent back to the medical oncologist's provider, the difference between one week for a liquid biopsy and potentially up to one month for a tumor biopsy NGS, that makes a big difference when we're talking about treating our patients. So turnaround time is very important. Liquid biopsy also potentially addresses issues of tumor heterogeneity. So instead of just taking a piece of one tumor site, presumably there are multiple tumor sites that are shedding DNA into the bloodstream that we can actually assess more globally the genomic changes associated with the tumor. And I'll show you one example. We can actually use cell-free DNA for disease monitoring as a complementary approach to looking at CT scans. So I would encourage everyone to think of liquid biopsy and tumor biopsies not as one replacing the other, but as absolutely being mutually exclusive. And I think most large academic centers, including ours, do these simultaneously right now, where we order a tumor biopsy, tumor NGS, and we order a liquid biopsy NGS at the same time. And they give us different and complementary pieces of information. So of course the tumor biopsy gives us key pathological information. What is the histology of the tumor? Is it adenocarcinoma, is it squamous, is it small cell? Some idea, is it mucinous or not? It gives us some idea of PD-L1 expression. The disadvantages to tumor biopsy from a medical oncology perspective, when we talk about biomarker testing is that turnaround time. So I'm often seeing the patient and then waiting three to four weeks before saying, here's how I'm gonna treat you because we don't have your molecular profiling of your tumor yet. Liquid biopsy, on the other hand, multiple studies have shown that there's a very high concordance rate, upwards of 90%, between the genomic variants that you detect in the tumor and the genomic variants that you detect in the blood. It is easy to sample blood. We do it all the time in our clinic anyway. So it's just another one tube that we get to send the liquid biopsy. The turnaround time is quite quick. But I would say, of course, the liquid biopsy cannot currently assess tissue histology. So some current clinical applications of liquid biopsy. And if you're interested, there's actually an entire conference on liquid biopsy ongoing in Miami later this week. So there've been multiple studies across multiple tumor types, not just lung cancer, but colon cancer, breast cancer, pick your favorite cancer. There's likely been studies, prospective clinical trials, looking at the utilization of ctDNA in that disease, and not just in advanced stage disease, but also in early stage disease as well. And at present, ctDNA is used as a basis for diagnosis to help us make our therapeutic decisions and to identify and track specific genomic alterations during the patient's disease course. So here's an example. This is a patient with metastatic colon cancer, a liquid biopsy is sent. The results show that there's multiple different genomic variants that are found in this tumor, a KRAS G12C mutation, a KRAS G13D mutation, and a BRAF V600E mutation. These three variants are used to help us select for and against specific therapies that we use for our patients in the clinic. Here's another example. And this is actually one of my patients. So this is a patient with advanced ALK-positive lung cancer. So ALK is one of the multiple different genomic variants we test for in all of our patients with lung cancer now. There are six different FDA-approved anti-ALK drugs. This is a patient who had advanced ALK-mutated lung adenocarcinoma who went on a clinical trial of a drug called nSartanib. And if you see here, and hopefully you can see my pointer, the x-axis here is the cycle. So each cycle of drug is one month. The y-axis is the allelic frequency, so the frequency in which a specific genomic variant is detected in the blood. And if we look here, and there's multiple different bars, hopefully you can see them from where you're sitting. The patient had what's called an ALK-L1152V mutation present at the start of therapy. So that is shown here. The L1152V is green, and you see this green at the start of drug. When the patient went on the study drug, that mutation went away. So here's cycle two, day one, where the patient comes in for assessment. We have another liquid biopsy as part of this was a clinical trial. And that mutation, that characteristic mutation now is gone. At cycle three, day one, we see additional changes in the blood. Here at cycle four, day one, we see that the mutation shown in green, the ALK-L1152V, is still at 0%. But we start seeing emergence of a different mutation shown here in blue, ALK-E1210K. So this is an emerging resistance mutation for this drug nSartanib. At this time, the patient is still showing a continued radiographic response by resist criteria. So already starting to see emergence of resistance. By cycle five, day one, you see a further increase in this specific resistance mutation shown in blue. And the patient is now having evidence of radiographic progression of disease. So this is one way we use liquid biopsy both to diagnose, to find the mutation in the first place, and ultimately track the disease course over time as the patient responds or does not respond to therapy. So multiple benefits of a liquid biopsy. Again, faster turnaround time for required molecular studies. As oncologists, we are graded on the percentage of patients who get molecular profiling. And so to us, because most importantly, this is how we treat our patients, but it is also a key metric for how we're caring for our patients. So turnaround time and accessibility of these molecular studies is absolutely critical. We can actually do a liquid biopsy from our own clinic. We have phlebotomy labs that blood is drawn and sent to where the tumor DNA is gonna be sequenced. It's less or minimally invasive, can address heterogeneity across multiple disease sites, and is quite practical for disease monitoring. But liquid biopsy is not perfect, and it is very much evolving. Huge topics, multiple fantastic reviews, and I've referenced several of them throughout my presentation. Some of the challenges of liquid biopsy. First, the technological challenges. This is not a simple test to do. You're truly looking for a needle in a haystack. And so the tumor fragments in the blood may be present at 1,100th of the normal blood fractions, the normal variants found in blood. So you are really looking for that needle. These, in the haystack, these tests need to be super sensitive. The half-life of cell-free DNA is about 16 minutes to two hours. So there's very rapid turnover of these variants. So it's dependent on continual shed of the DNA into the blood. And non-mutant fragments, so that DNA that we're all shedding right now as we sit here just from normal body turnover, are the most abundant. So the circulating tumor DNA is actually the minority of circulating free DNA in the blood. There are biological challenges associated with liquid biopsy. Not every tumor sheds DNA. So I certainly have patients with metastatic disease who I send a liquid biopsy and don't detect anything. The biology of actual tumor DNA shed, we really don't understand at this point, but lots of active study on this. Clonal hematopoiesis can complicate the interpretation of liquid biopsy. So we all know as we age, we just tend to gain mutations in our hematopoietic cells. That has to be subtracted out from liquid biopsy to really understand the contribution of the tumor versus the contribution of other potential disease, pathogenesis, in the blood. And what is the right number of genes to evaluate? So these tests have to be very sensitive. That means very deep sequencing. So for example, when we do tissue DNA sequencing, it's usually at about 50x level. When you do blood NGS, it's usually around 1,000x. So you're sequencing the same fragment 1,000 times. Reporting and interpretation. So the variant calling, actually saying, is this mutation a true mutation? Is it a variant of uncertain significance? It is a true pathological mutation, is not trivial. And how these complex domic reports are actually presented to physicians is not trivial as well. And these are typically PDF documents that we're getting faxed to us. They're often not readable, to be honest. And so integration of these results into the medical record is something we all struggle with in our clinical practices. And finally, implementation into the clinic. I would be remiss if I didn't say there are lots of regulatory issues around this. Who approves CT DNA tests to make sure that they're sensitive and specific? How are they paid for? Fortunately, insurance. I've actually never had a patient be denied a liquid biopsy. And who orders CT DNA testing? Who takes responsibility for those results during a patient's disease course? So despite these challenges, liquid biopsy is here to stay, in my opinion. And there's multiple future uses of liquid biopsy. So let me articulate a few of those before we close. Additional clinical applications of liquid biopsy. Right now, we are using it primarily for finding variants of known significance. So find a variant, it's that variant is tagged to a specific FDA-approved targeted therapy, and then that therapy is given to our patients. But that's not the only way we can use liquid biopsy. There are huge investments right now, and some of them are actually here today, like Delphi, into cancer screening. Can we use cell-free DNA to actually identify cancer before we find it radiographically? Can we use liquid biopsy to assess minimal residual disease? For example, in patients who have resected lung cancer, can we use liquid biopsy to actually find the disease recurrence before it's visible on a CT scan? And then, of course, treatment selection. There are multiple different analytes in the blood. So while predominantly we're talking about CT DNA right now, in the future, we'll be thinking about other analytes as well. Another huge area of study. So some of those analytes, cell-free RNA. It's a little bit more complicated to actually extract and sequence cell-free RNA, but possible. That gives us different levels of information about the expression of different variants across tumor types. We can use exosomes or extracellular vesicles. These are little pieces of tumor that are shed into the blood as membrane-bound pieces of the tumor itself. And then, of course, circulating tumor cells, which are present in high abundance in tumors like small cell lung cancer. Beyond genomic variant calling. So right now, we're largely using liquid biopsy to say, do you have an ALK mutation? Do you have a KRAS mutation? But we can actually learn a lot more from even the sequencing that we're doing right now and a few of those areas where you'll see liquid biopsy expanding in the future. A big one right now, fragmentomics. So this is the study of the actual size of the pieces of DNA that are being shed into the blood. And that actually gives us information about the origin of the tumor and the active transcription and translation that are ongoing in that tumor. Methylomics, so looking at epigenetic changes in the tumor DNA that may give us an idea of the pathobiology of the tumor. And nucleosomics, an idea or a study of the nucleosomal positioning across the genome in the tumor. And last but not least, we largely use liquid biopsy now in plasma, but there are other body fluids we can use to actually identify genomic variants associated with the tumor. Anything from CSF to saliva to pleural fusions, urine. These are all areas I think in the future you'll see significant growth in our utilization of liquid biopsies moving forward. So I end with my title, will liquid biopsy replace pulmonologist? Absolutely not, these are complementary studies. They are not mutually exclusive. We love them as medical oncologists because it gives us information fast, but we recognize that it is part of the piece of the puzzle for our patients, but certainly not the entire puzzle. And with that, I'd like to give a huge shout out to the interventional pulmonologist, Vanderbilt. So here's Otis, Fabian, Siwe, who is our fellow, who is here at the conference. She's lovely. And Rob Lentz, they are awesome. It truly takes a village to treat our patients with cancer, and none of us can do this alone. So appreciate all your collaboration. I'd love to chat with anyone afterwards. Thank you again so much for having me, and please feel free to contact me anytime. Thank you.

lung cancer biomarkers

biopsies

molecular testing

liquid biopsy

genomic analysis

tumor heterogeneity

PD-L1

immunotherapy response

disease monitoring