Hi, everyone. I'm Aaron Farberg, a board-certified dermatologist based in Dallas, Texas, and welcome, again, to another chapter here in the skin cancer section brought to you by the AAD. The section I'm going to be talking about is advances in diagnostic and prognostic testing in cutaneous melanoma. Again, I'm Aaron Farberg, based in Dallas, Texas. I have a private practice for dermatology, and I'm also affiliated with systems, including our own dermatology residency program, the Lake Granbury Medical Center Residency Program. In any case, let's get started. That's me, in case you wanted a second picture. Here are my conflicts of interest. I am an investigator as well as advisor for Castle Biosciences, which is the maker of a couple of the tests that we will be talking about today. Here are the objectives. We want to describe the challenges of accurately diagnosing, first just simply diagnosing, and then predicting the behavior of malignant melanoma. We also want to compare GEP, gene expression profiling technologies, and how they can predict the behavior of melanoma. We want to evaluate the emerging data surrounding the outcomes of utilizing these technologies, as well as then briefly outline some of the guideline recommendations, as well as the implementation of these gene expression profiling technologies. Let's first talk about the challenge here, which is simply diagnosing malignant melanoma. Of course, if you've been reading any of the news, you know that there's always some sort of controversy about diagnosing melanomas. Is it too many? Is it too few? Of course, I would say, which ones do you want me to leave on a patient? Of course, the answer is none, but for the most part, we're really good at finding these. We biopsy a lot of patients, and we do a lot of biopsies of suspicious pigmented lesions, and for the most part, we know what they are. They're either benign or they're malignant, but there's always this group that sort of falls in that, well, I'm not totally sure, and those numbers range. Here you see numbers ranging between 15%, 37%. We could argue about those, and I think, of course, a lot of it comes down to the experience and expertise of your dermatopathologist, but the point is, is even if it was a couple percent, that's a concerning number. More importantly is, what are people doing with these lesions? The answer to that is most of them are treating them as the most malignant diagnosis it could be, so if melanoma was within your differential, and it comes back as this ambiguous melanocytic lesion, odds are most people are treating it as a melanoma, so again, the question would be is, are we doing these patients the right service, or should we be considering different sorts of workup? This is what we're going through now. Again, you see a suspicious lesion. You can either monitor it or biopsy it. If you biopsy it, you get a pathology report. Now sometimes you get a ambiguous lesion, or you get a clinical pathologic mismatch. Let's say you thought it was malignant, you're, you know, you're betting, you know, your money that this was going to be a malignant melanoma, and it comes back as a benign lesion. Hmm, you know, that's a serious mismatch. What is your plan going to be? In these cases, when you get these ambiguous lesions, oftentimes your dermatopathologist has already met with their friends or their group, or they have a consensus group that they meet with, you know, in the mornings, or they'll also run some ancillary testing. Of that, many of you are familiar with things like FISH or NGS or CGH, all these acronyms for testing these melanocytic lesions to really determine the malignant potential of these lesions. You know, of course, many of those are very expensive. They're only found in large academic centers, where, you know, again, cost can sometimes be very, very expensive. But there's another option, which we want to talk about tonight, which is, or today, which is the topic, is gene expression profiling, and how that more affordable type of testing can lead you to a much more definitive diagnosis. So how do GEP technologies help with this better diagnosis? Well, just taking a step back, going back to biology in your high school class, of course, you've got DNA, which is the instructions for pretty much everything that goes on in our body. But, of course, it includes a lot of instructions which aren't involved. And that's where we have the RNA, which is really the specific, hey, this is the type of proteins that we want you to build, and how you're going to build them. This is, you know, perhaps a little bit more accurate or reflective of what's going on in the body. But then, you know, you could also say, well, let's just look at the proteins, and the proteins are the actual built, after you listen to those instructions, what you've built. There's a variety of ways of looking at this, and one is neither right nor wrong. They're all sort of interesting. They just have required different techniques that require a variety of different laboratory technologies, which come with time and also come with cost. But for gene expression profiling, I think it's sort of that happy medium, you know, the porridge isn't too hot or too cold. And it's quick, and it's also affordable, and gets you a good understanding of exactly what's going on with a cancer. Again, this isn't necessarily new technology. This has been around for many years, and is used in a variety of other cancers as well. So in any case, when you're looking at, you know, genes, you're not just looking, are they on or off? You're also looking at their overall expression. How turned on are they, or turned off are they? How does GEP testing really work in a laboratory? So essentially, the lab will get the slides from the original block of tissue, you know, for this melanoma. A dermatopathologist should be reviewing those slides and circling the actual tumor tissue. That tumor tissue is then oftentimes macro-dissected off, and of course, the paraffin is removed. The RNA is extracted. It's reverse transcribed into cDNA. That is then quantified by a qualitative PCR. And then once you have all the genes in question that you're interested in, an algorithm will be performed or applied, and that algorithm, again, is looking at the relationship of all these different gene expressions, also in relation to each other. And then from there, it should have an output where the result should impact your clinical decision-making. So, again, where would this GEP testing lie? It's these ambiguous lesions, your dermatopathologist just is not sure, you're not totally sure, and so you can get this additional information in, you take it in collaboration with the actual clinical data that you have to come up with a diagnosis. Again, remember, you as the clinician are actually making the diagnosis. It's not necessarily just what's written on a pathology report. So the test that we want to discuss here tonight, the diagnostic testing is the 23-GEP used. Here are the various genes. If you want to get into the weeds, oftentimes people ask, well, what are the genes? Are they related to melanoma? The algorithm, it goes through that, and it classifies the lesion as either malignant, benign, or intermediate. Now, there's also a lot of good data on this test, the 23-GEP test, and again, I'll just use the trade name of it so you at least know what it is. It's the MyPath test from Castle Biosciences, but it's 23 genes, which is why it's a 23-GEP. There's a lot of studies out there for it, and I just want to highlight an important one. This is a prospective study of 25 patients that were followed out 38 months, so a solid follow-up period time on average, and these were diagnostically ambiguous lesions. The GEP results were then included with the clinician's decision-making process, and that included, do you excise it or do you not re-excise it? After following the patients for a good amount of time, you can see that there were no recurrences, no metastases reported in either of those patients. In this case, following the GEP testing led to good results, saved the patients an unnecessary procedure, a re-excision. Now, again, would I want to see 1,000 patients? Of course. Would I love to see it for 10 years? Of course. But, again, this is good data, prospective data that at least gives us some support that this testing is very worthwhile. Don't just trust me. There's a number of guidelines reviewing this, including all the ones you see here that include this type of testing, along with the other tests that I had mentioned. This is a real concept that when you get to an ambiguous lesion, you shouldn't just go ahead and just call it a melanoma and then give somebody that diagnosis and treat it as such. There is value in actually identifying the real malignant potential of these lesions. Now, what's also interesting is that once you do have a malignant melanoma, this testing can be reflexed if it's invasive into a prognostic test, which is the 31 GEP test looking at melanoma prognosis. And the advantage here is that you don't need additional tissue. You don't need additional slides. That's important because we all know that this tissue of these melanomas is extremely valuable and there's not always a lot of it. And so we want to take extra good care of it. So since we're on that topic, let's now move on and talk about melanoma prognosis. Taking a step back, when you are diagnosed with a melanoma, at least the way I see it, is that there's a couple of questions that I want to answer. First, do I need to send this patient to a surgical oncologist to have a discussion about a sentinel node biopsy procedure? And to answer that, I have to understand, what is my patient's risk of having a positive sentinel node? Then the other question I ask is, well, what's their chance of this coming back? What's this chance of their melanoma killing them? That if it's high risk, well, obviously I'm going to do a high intensity surveillance. And if it's low risk, well, then maybe we can forego some of that. Essentially, what we want to do is we want to align our treatment, our management, our surveillance of these patients. If we're going to include immunotherapy, if we're going to include imaging, and we're going to be very intense with it, we need to align that intensity with the actual patient's risk. This is to conserve resources, but it's also to treat the patients in the best way possible. If they're truly low risk, then I assure you they don't want to go through any unnecessary procedures. So what are we doing now? Well, we've got staging, TNM. We've got AJCC. Here we go. And you can see, for the most part, when you have these early stage, stage 1a, 2a, these thin melanomas, essentially they tell you, well, you should see the patient back every 6 to 12 months. I almost guarantee most of you aren't doing that. You're seeing them every 3 months, right? That's what I know most of my colleagues are doing. At least for the first couple of years. And then it's imaging for only as indicated thereafter. So again, it's cut it out and hope that everything is all good. And oftentimes, most of the time, that's the right thing to do. When it's a thick melanoma, well, then you're going to be seeing the patient back a little bit more, and you're still going to consider imaging for any recurrence or any sort of symptomology. But there's otherwise no reason to consider routine imaging for screening in this sense. So let's move on and start off with, let's talk about senolibno biopsy. Who do we recommend a senolibno biopsy to? It's pretty straightforward. We mostly do it by T-staging here. If they're T1a, their risk of having a positive node is less than 5%. If it's a T1a with some high-risk features or a T1b, then it's between 5 and 10%, and you should consider it. And if it's anything deeper than that, well, then it's greater than 10%, and they are eligible, and you should be offering this. Where do these numbers come from? Of course, they sound like very nice round numbers, 5, 10, but they actually come from big clinical studies, MSLT, one that really determined the risk of a false positive in the senolibno biopsy procedure itself. And so, you know, you wouldn't be doing a procedure if the risk was any more than that. So that's where the 5% is coming from here. But as great as our staging is, there are inherent limitations to it. And we know that, and we agree to that, and we're okay with that in one sense. The point is, is that you're going to have patients that you otherwise think are low-risk, and they go off to have bad outcomes. Or the opposite is true, where, hey, you think, actually, they're very high-risk, but they go on and live a totally fine life. Now, what's really concerning is that of the patients that are otherwise considered low-risk, stage 1 and stage 2, we should have really good outcomes with that. More than half of the deaths from melanoma are from that group, from the patients that we said, oh, it's really no big deal. Again, they were originally staged as stage 1 and stage 2. Most of the deaths are occurring from that group. Now, yes, that's also just due to the sheer numbers. But the point is, is if you were in this group, you'd want to know, are you truly in this group, or would you actually be requiring some additional management that could actually help you? And so, again, it's really important, even though we utilize staging in a binning way, we also want to be treating our patients in an individual way. And that's where the 31-GEP comes in. The 31-GEP, it's actually, to be more specific, it's this integrated 31-GEP, because the way it works is it takes these clinical pathologic features, and again, depending upon the test, it'll bring in Breslow thickness, ulceration, mitotic rate, or age. It will include the GEP score, and then it will be able to tell a patient's actual individual risk of having a positive note. Now, for risk of recurrence, it utilizes a few other different clinical pathologic features. Now, in the development of this test, a number of clinical pathologic features were looked at. So if yours is not there, don't panic. It's not that it wasn't looked at. It just didn't move the mark statistically. So these are the important factors. And again, we're already including many of these in other sort of staging systems. So it's good to be able to utilize all of this together to get a more precise prediction for your patient's risk. Now, this is what's kind of cool, is thinking, okay, to figure out a sentinel node, that 5% positivity risk, what does that really look? What does that mean? Well, it means that for every 20 patients who are otherwise eligible for sentinel node, if you did not perform a sentinel node, then 19 patients would have had a negative node, and you would have been right. Good, you didn't do the sentinel node. But you would have missed one and so what that's telling us is that we're okay not doing these nodes on all these people and we're okay with missing one in every 20 of these. So it's a 19 to 1 true to false negative ratio which is at that 5% risk. This is currently at the NCCN recommendation. We're cool with missing that one patient and that has just been what we're saying right now. So the point being is that if we're going to do anything else we better be doing better than that and so let's take a look at how the 31 GEP works. So you've got the test, it's got the clinical pathologic features, mixes with the GEP and you get the patient's precise risk for a positive node. Look at the actual metrics, the accuracy metrics you see. It's got a very nice high NPV and you see all the other factors there just so you know the paper that's available that describes all that is here so you can snapshot that. But before we move on there's this other test. It's the 8 GEP plus clinical pathologic features. This new assay is currently under investigation by a number of academic institutions which are invested in it and it's been developed to identify exactly this one and only one question of what's the chance of having a positive node? And it looks at eight different genes and it also includes a couple of different clinical pathologic features. Sound familiar? But it doesn't exactly separate them out. It actually includes them all and you can see that if you just take the clinical pathologic features versus the GEP model they do well but when you combine them it does even better. The hard thing here is that it's difficult to say exactly what is providing that lift. So in any case going back to that 19 to 1 true to false ratio that's what we're currently doing now. Again we're okay not doing nodes on these people and missing one for every 19 that we otherwise would have been right about. And let's look at these two different tests head-to-head. We've got the 8 GEP and the 31 GEP and looking at the the different studies and again this was looking at the U.S. study that it was looked at for the 8 GEP. Most of their other studies are international. So looking at them at their absolute best you can see that their ratio was actually 17 to 1 when it came to their negative versus true negative there. So again it wasn't quite beating what we're currently able to do with the NCCN guideline. When you look at the I31 GEP there you can see the true to false negative is actually 30 to 1. So there is an improvement there above and beyond what is currently being done. But you know again don't just trust this study. Also this is another study that was done by Dr. Marchetti who's at Memorial Sloan Kettering. He has been a very strong critic and advocate of all these types of technologies and he developed his own utility model analyzing this integrated 31 GEP. Wanted to know if it worked and what you see here is anywhere it's positive that shows according to him this relative clinical utility. Interesting the T1A with high-risk fixtures didn't show it and again you don't have to have it's not a delta. Anytime there's positivity it's showing that yes there is absolute utility. What would be interesting is if they did this sort of study for the 8 GEP which unfortunately they did not. Moving on now we will have to understand what is this patient's risk of recurrence. Why we care about that? Well we want to know hey is it is it a high risk or is it a low risk? Should we be watching this patient very carefully or not? And what's very exciting about GEP technology is that it takes what otherwise here in the black dots is what our current staging our AGCC staging shows us for melanoma specific survival, recurrence free survival, and distant metastasis free survival. And again we could split this out by T staging it's just to give you a concept of the personalized medicine that that the GEP testing is able to do. That's the bin stage and if you were to include the i31 GEP you can actually see each individual patient's risk and you could see it fans out showing you hey for the most part you're right that staging is right it bins it but once in a while you get these that essentially cross entire staging lines and so you know what you otherwise thought was a stage one actually is maybe behaving like a stage two with in regards to their risk or even worse you have a stage two that's actually acting like a stage three and we all know we treat these patients in a very different way. Another thing you want to see is repetition you want to see lots of studies not one not two not three and you don't want to see a bunch of international you want to see what's going on in your patient population with the ones that are going with your patients and needless to say there's a number of papers including multiple meta-analyses including prospective studies that show that this the 31 GEP is able to stratify these patients. All right if you've made it this far now you got to pay attention. All right here is a very important study that was done linking the 31 GEP with the SEER database this is with the National Cancer Institute. Of course this test has been reimbursed by CMS our government and so you know the government what they do best is they want to know are they getting their money's worth and so little did anyone know that you know the government said well hey look we've got data on patients that had been receiving the 31 GEP and we want to know you know are we getting our money's worth is it doing what it was you know what all these other studies have said it was doing and so you know again just to highlight what is a SEER database we don't have socialized medicine we're not one of those countries thank goodness but in any regard we do have this thing called the SEER database which is supposed to be the most representative form of that includes a number of different states and people are actually paid these large academic centers included to go out and gather data so it's not just hoping that we report it because I know you all do but not everyone is good at it they actually go out and get all this data on certain cancers and melanoma is one of them and so the the company that has the 31 GEP worked with the National Cancer Institute to essentially link the patient's results with their actual outcomes but they went on beyond that and they thought well okay if we have patients that got the test can we compare them to a very similar group that didn't get the test and see if there was a difference in outcomes and so we're going to talk about exactly that let's first talk about does the test work does it stratify patients like we thought it would and you know again they looked at diagnosis from 2016 to 2018 because you know there could be different access to adjuvant therapies wanted to just simply say you know does this test stratify the other thing they looked at when they wanted to compare the two groups the patients that got the test to the ones that didn't they wanted to find two groups to compare that were otherwise similar in every way possible and these are all the different covariates that were looked at you can see them here and you might say well oh you're missing one or two they're not this is literally every covariate that is recorded by the seer database and this is what I mean by this is the best that you could possibly ask for or hope for or want you know this is this is the American Cancer Institute at its best national NCI but you know from America and without any further ado here you have the result you know it's great you see this that it actually stratifies patients it looks like just like those other Kaplan-Meier curves and so it's it's reassuring that looking at this prospective data unselected group not you know company-based which is also very important this is literally this is data from the National Cancer Institute they're the ones that ran all this and you can see that it stratifies and look at the sample size you know this is a serious group here also important upon multivariate analysis you know what is perhaps the best predictor of melanoma specific survival and overall survival again you can see here it's the GEP score it's not to say Breslow thickness is an important or ulceration or lymph node status is an important it it's simply saying that it that the 31 GEP was a the most important predictor and it was also independent that's also what I talked to you about earlier is having a a test that provides independent information not just you know a data that goes along with all the other clinical pathology features we have now again we want to get to the exciting part so here we have a group of patients that had the test and compared them three to one to patients that didn't get the 31 GEP we wanted to know how did these patients do and the answer is you were 27 percent more likely to live if you got the 31 GEP test overall survival that was from melanoma overall survival was a 21 percent improvement more likely to live if you had the 31 GEP now what's the conclusion of this study are you going to say that oh a test is what makes people live longer of course not that's that would be silly but if you look back through all the data there's a number of chain links there showing that it modifies physician behavior that there's those behaviors are well documented that behavior may be different across different clinicians a dermatologist may act differently on this than a medical oncologist differently from a surgical oncologist but we know that that there's utility there those changes in management then lead to this level of improved outcomes for patients now just to give you an idea of reference you know there's another test that is widely used in breast cancer the oncotype dx breast cancer test and to give you an idea that test improved survival by 0.5 which you know again you might think oh no big deal but no that these that level of change is huge over the course of a population looking at the 31 GEP the absolute mortality difference is an even greater improvement 1.1 and that's just at three years now again you know this this data set is only at three years out so I don't disagree with you I look forward to seeing this at five years but at least it's in it's all been in the right direction and continues to be now here's an exciting story that I was very excited to see I mean this is a good study showing you know what are some of the changes what are some of the things that can be done by us us clinicians that can actually lead to that improved overall outcomes and the answer is imaging so what this study did and it was done at two very large well-renowned academic centers it's a retrospectively identified sentinel node negative melanoma patients and they were put into two different groups an experimental group that if they had the 31 GEP and it was a 2a or 2b a high risk result they got routine imaging hey we're just going to image these patients they weren't even specific about the imaging but hey you're going to get imaging every 6 to 12 months you know whether you need it or not and we talked about what some of the other guidelines or suggestions have told us saying well you only should image if needed if there's an actual symptom and that was what the control group is so we're going to just do that for the control group and compare it to hey look the 31 GEP identified a high risk group let's just image them and see what happens and of course you know those that you know had a metastasis or recurrence were then offered checkpoint inhibition advanced therapies and just to give you a look at these different groups because some people might say oh well you know maybe it was ordered on you know a unique subset of patients or there was something unique between the groups absolutely not in fact you know the data was you know if anything went the other direction in regards to the the the test or the the group that was tested and got the two a's and two b's you know they were already a little bit higher risk and in any regard what you see here is that the group that underwent the imaging that were found to be high risk by GEP and then underwent imaging they actually found the tumor burden at a much there was those patients had much less tumor burden by a significant amount so in other words they found the cancer while it was smaller also they found it sooner so with melanoma when you use GEP and you find that that these patients are higher risk you can do imaging on a consistent basis and we know here that you're going to find these that this melanoma any sort of recurrence or metastasis sooner while it's smaller can't repeat that enough that's like the kicker now maybe 15 20 years ago when we didn't have immunotherapies that didn't matter i don't disagree it you know maybe it wouldn't have mattered but we now have other very good evidence to show that with the current treatments that we have when we find melanoma metastatic melanoma sooner and when we find it while it's smaller it actually leads to better outcomes so there is a reason to do all this now and this is what has led you know as i've said the different links of evidence this is what has led to that improved outcome for patients that have had GEP testing so again if you find that the patient is high risk even though you've already done a sentinel node and you know it's negative because that's usually what it is you could still identify a riskier group within that that node negative group and image them on a routine basis every six months and find any sort of occurrence sooner real quick some of the other recommendations regarding the implementation of the GEP this is the skin cancer prevention working group which i am a part of along with a number of fellows from the very famous Dr. Daryl Regal who invented the ABCDs of melanoma and has devoted his entire i want to say life but at least his career to melanoma and pretty much only melanoma now and it's just showing that you know there are limitations although the NCCN and AJCC are incredibly important you know there are limitations in in just simply staging and binning and we're all moving towards a precision medicine future and GEP has been a validated reproducible and consistent tool in identifying our patient's actual risk and we all know now that it can actually impact our patients outcomes finally what also might be important what patients think yes that really does matter you know it and i know it there was surveys that were done of patients regarding GEP testing and you know what influenced you know how they were feeling what their decision making was and you know for the most part here 66 of the respondents indicated that the results were useful or extremely useful so again you know this is highlighting that this also helps patients and their from their perspective as well and with that i'm going to conclude i know it was a long journey we've been on i really really appreciate all your attention and look forward to seeing you at one of the AAD meetings or one of the local meetings

Dr. Aaron Farberg

dermatologist

Dallas, Texas

cutaneous melanoma

gene expression profiling

23-GEP test

31-GEP test