Hi, I'm going to be talking today about molecular tests for the diagnosis and prognosis of melanoma. I'm Laura Farris, and I am a dermatologist at the University of Pittsburgh. So first of all, I'm going to talk about gene expression profiling. So let's start with what is it. This is looking at RNA expression within a tumor, and the profile or gene expression profile gives us an idea if a lesion is behaving more like a benign or a malignant lesion, or if it's more or less aggressive. So here, we're looking at the relative expression of two genes, one shown in yellow, one shown in blue. And if the predominant gene that is expressed is blue greater than yellow, that is likely something that has more benign behavior. Whereas if the yellow gene is predominantly expressed relative to the blue one, that would be a lesion that is going to have potentially more malignant behavior. So in addition to talking about what GEP is, let's talk about what it's not and how it differs from other genetic tests, the first one being germline mutations. So we know that germline mutations are heritable, and that mutations in genes like CDKN2A or BAP1 confer an increased risk overall of developing melanoma within a patient. So these are DNA mutations carried in every cell in the body in a patient. And then there's also acquired mutations. So these are not inherited. They're not found in every cell. They are mutations that can develop within a melanoma, the most common one being BRAF mutations. About 50% of melanomas will have a BRAF V600 mutation. And so this is a mutation found in the DNA, but it is acquired. And so a patient could have one melanoma that has a BRAF mutation and one that does not, for example, and it is not inherited. So let's talk about how GEP can be applied in melanoma. So one is making a biopsy decision for pigmented lesions. So there's something called a pigmented lesion assay. It is a piece of adhesive plastic that goes over a lesion, non-invasively collects genetic material and can be used to help for us to make a decision of, is this suspicious enough to warrant a biopsy? Why do we need that one? If we look at the top panel of lesions, we can see that melanoma and non-melanoma can clinically look very similar to each other. And on the bottom panel, we can see that these are both melanomas. The one on the left is a pretty obvious melanoma. We don't need help with it. The one on the right is actually an early melanoma as well. That is one that's a little bit more difficult to diagnose clinically. To determine how good we are at making a biopsy decision using our naked eye exam or naked eye plus dermoscopy, we can look to reader studies to say, what's our sensitivity? How often do we accurately say that looks like melanoma or I should do a biopsy? It's about 63 to 97% depending on the study that you look at, which is going to depend on the composition of the lesions in that data set. We can also look at number needed to biopsy. So how many benign lesions do we biopsy for every one melanoma? It can vary by experience level, but on average, it's about 13 for US dermatology practitioners. Now GEP is essentially based on the idea that genomics are really what drive melanoma progression. So it is genetic changes that drive cells to proliferate, to lose their normal differentiation and ultimately form cancer. So what we are looking at clinically are those macroscopic changes when we biopsy. We try to confirm that with microscopic changes. But if it's really genetic changes that drive it, the thought would be that then if we look at the genetics or genomics of a lesion, that should really tell us how worried we should be about it. So in the pigmented lesion assay or this adhesive tape assay, there are three different genes that are evaluated. One is called LINC00518. This is a long non-coding RNA and this has been implicated in melanoma invasion and metastasis. Also in other cancers such as breast and lung cancer. The other RNA that's measured is PRAME, which stands for preferentially expressed antigen in melanoma. And so we know that if we look by immunohistochemistry, that melanoma is more likely to express PRAME protein than our benign nevi. And so this test actually looks for the RNA expression of PRAME. There is another gene that was included in the test and this was actually not RNA expression, but it was a DNA mutation specifically in the TERT promoter region. This is something that helps to regulate telomere length, which regulates cell mortality. We know that TERT promoter mutations are much more common in melanoma than in nevi. I mentioned this, however, this test actually no longer includes TERT promoter mutations for several reasons. One being that the getting enough DNA to measure it was challenging. The other being that there are many lesions that are, there are lesions on sun damaged skin that are non-melanocytic such as seborrheic keratosis or lentigines that can have TERT promoter mutations, but are not melanoma. So looking at the pigmented lesion assay, this is data used in the validation study. Looking at just the two gene tests that the RNA expression of LINC or PRAME, if one or both are present, considering that to be a positive test and neither are present as a negative test, the sensitivity of the test is 91%. The specificity in one prospective study was 69%. If we add in that TERT promoter mutation, the reason that was done was that it did increase the sensitivity to 97%, but we can also see that that slightly lowered the specificity to 53% as well. Okay, so this is an example of this test, the adhesive patch that we can see here. So you know, in real world use, you would take this test, if it were positive, presumably you would bring the patient back for biopsy. If it were negative, you would not. But what about those negative PLA lesions? How comfortable can we be with a negative test to say that is not a melanoma? And so this is a study that was done in 2021 following patients who had a negative PLA test. They had 1,781 patients, 69% of them had clinical follow-up. Among those patients, 10 or 0.8% of them actually had that lesion diagnosed within the next year as being a melanoma. These were early melanomas in Site 2 or Stage 1. Also a subset, 304 patients who had negative tests were retested 6 to 12 months later. Among those tests, 34 came back positive on retest, so about 11% retest were positive. These were all biopsied. Of those 34, 3 or about 1% were found to be melanoma. They were all melanomas in Site 2. So this really kind of suggests that the negative predictive value of this test is around 99%. Okay, these are some examples of cases in which I've used this test. This is a patient who is 67, no history of skin cancer, wanted this lesion on her nose, treated cosmetically. The plastic surgeon asked for clearance from dermatology. We talked about, you know, my clinical impression, then we also talked about this test. And so we did do the pigmented lesion assay. This lesion expressed neither link nor frame, so risk status was green, she was free to go ahead and treat the lesion. This is another patient who is 38 years old, had a history of melanoma in Site 2, had several nevi. She'd had 10 biopsies in the past 13 months, was kind of experiencing biopsy fatigue. I did see this lesion I'm pointing to in this picture and said, I'm not, I think it looks atypical, and I would like to biopsy. She said, is there any way that we can avoid that? What else could we do? So we talked about this assay. We did this test, and this was link positive. And when I did biopsy based on that result, which she was then happy to have the biopsy, it did come back as an early evolving melanoma in Site 2. Here's another case of 85-year-old man, new lesion on the cheek, cosmetically sensitive area. We did order this test, in this case the lesion expressed frame but not link. This was a case where we did a TERT promoter mutation data, and it was negative. And then on biopsy, this was a lentigo malignant 0.5 millimeters. Now these are examples where this test was helpful to me. There are obviously pros and cons to most tests, and there are cons to this one. This was a study that was published where they looked at the real world use of this test, and what they found was that there were a high number of tests that didn't return a result because of insufficient material to run the full assay. This happened with 12.5% did not have sufficient quantity to run the assay at all. Interestingly, over 70% of the samples from this practice had a quantity not sufficient to run TERT promoter mutations at all. They also found that there was discordance between the pathology result and the tests about 38% of the time. So either the test was benign, but the pathology showed melanoma, or the test was positive, but the lesion was found to be benign anyway. This does require doing the test, and then if it's positive, bringing the patient back at another day, and it can take, so that's one issue. The other thing is that the authors noted that it did take extra time to do and document the test. It wasn't really a time saver in their hands, and that the cost of the test was two to three times that of doing a biopsy and getting a pathology. Another way to think about this test is that the genomic positivity rates, and their sort of positive and negative predictive value, are really going to depend on how you use the test and the setting in which you use it. So one of the validation studies that I mentioned of the two-gene test was performed in two pigmented lesion clinics. So these are specialized clinics with physicians who generally see high-risk melanoma patients, and patients who generally have had a history of melanoma. And in this case, the test was performed on any lesion in which the evaluating dermatologist decided that a biopsy should be performed. All lesions were biopsied, and for which the pretest probability of melanoma was determined to be between 1% and 99%. And so in this series of 203 lesions, 27% of tests came back positive. Of those positive tests, 35% of those tests, of those lesions, were actually melanomas. So I look at this as a way of saying, my plan is to biopsy this. I would use this test result to say, do I really have to biopsy this? But this test could also be used, is also used in real-world use. And so looking at registry studies which capture real-world use, I'm going to present two. Registry one just included the two-gene test. So the setting here were general derm offices with sort of all-comer patients. And so in this setting, 10% of tests were positive. Of those positive tests that were biopsied, 19% were melanoma. And in registry two, patients could either have the two-gene test or, in some cases when it was available, the TERT promoter mutation or the three-gene, including the DNA, mutation was added. And so in that case, 13% of tests were positive. And of those tests that were positive, 13% were melanoma. So I'm going to suggest that the way that this test was potentially used in the real world was to answer the question of, maybe I should biopsy this or maybe I should just wait and observe it. So maybe this was used on things that the doctor or the provider was sort of on the fence about biopsying. These were potentially lower-risk lesions, and we can see lower positivity rates. So important take-home point here, pretest probability is going to impact positivity rates. And one of the questions is, am I using this to avoid biopsy or am I using this to find things to biopsy that I maybe wouldn't have biopsied in sort of routine setting? And so wonder what the ways to think about what is the impact of these data and how do they sort of impact practices looking at likelihood ratios? So a likelihood ratio positive is sort of a rule-in test. So you're asking the question, with a positive test, how much more likely is it that this is actually going to be melanoma? And with the likelihood ratio negative, you're looking at a rule-out test. And so you're saying, if I get a negative result, how much less likely is it that this is going to be melanoma? So we can look at likelihood ratios for this test. I also looked at it for melafine, which is not clinically available now, but FDA-approved device for melanoma detection. So if we look for these different tests, we can see the likelihood ratio positive range from 1.3 to 2.9. So that means taking your pretest probability, if you had a positive test that was either 1.1 or 2.9 or 2.1 times more likely to actually be melanoma. The likelihood ratio negative, you can see, if you multiply that by that, that is going to cut your probability down to less than 20% or even lower, depending on which version of the test you're using. So post-test odds are basically calculated as pretest odds times the likelihood ratio. So what does that mean? If at baseline, you think the chance of this being melanoma is low, I'm going to say it's a 1% chance it's melanoma, and you get back a positive test and you've used 3-gene assay, you would say, well, all right, at baseline, really, it looked like it was a 1% chance. I guess I should now think of this as having a 2.1% chance. Is that going to sway your decision to biopsy or not? What if you thought, oh, there's maybe a 20% chance that's melanoma? And then you did the 3-gene assay, you got back a negative result. Then you would say, well, I guess there's really only a 1.2% chance that this is melanoma. And that might sway you not to biopsy. You could imagine if you thought there was a 100% chance that this was melanoma, and that went down to a 6% chance, that might be sort of a different calculation in your mind. So I think it's really important to think about what lesions you're using this test on. This was another study that looked at registry data and looked at how often was the test positive. Looking at the 2-gene test, we could see 9.5% of lesions were positive. Almost all of them were biopsied. And then of the positive cases, those 316, we can see the distribution. We can see that most were benign lesions, but that the melanoma in site 2 was more common than T1A lesion, which was more common than T2A lesion. So if we think about what this means cost-wise, we can say it's about $760 to run the test versus using Medicare payment data, about $175 to do a biopsy plus look at H&E histology. So if we look at this for 3,418 lesions, and we look at the 316 that were biopsied, and the fact that we found 47 melanomas in site 2, 11 T1As and one T2A. That means that the number needed to test to find any melanoma is 58. And so it's about $44,000 worth of testing to find one melanoma. The number needed to find an invasive melanoma is you got to test about 285 lesions. So that's about $216,000. And then if we look at just to find that one T2A among 3,418 tests, that was about $2.6 million. Okay. The other thing that's important to understand about a genetic test is that it is regulated differently than a device. So in this slide, you can see an electroimpedance spectroscopy device that is FDA approved for melanoma detection. We can see a device that's a smartphone app that, you know, gives feedback of whether or not the biopsy should be performed. Because those are both devices, they are under the regulatory oversight of the FDA, they must undergo full review, looking at the data, balancing risks and benefits, and be approved or not. By just sort of convention of how we do things from a regulatory aspect, gene expression tests or lab developed tests, they are under sort of CLIA oversight. But they are, which means that the test has to be run and reproducible, but there is not a separate body looking at the data and the impact. So that means really that our, you know, understanding the data is really critical for gene expression profiling tests. So how many validation studies are out there? So sort of highest level of evidence, prospectively collected lesions, where every single biopsy was read, was, or every single lesion was biopsied, every specimen was read by three dermatopathologists and used consensus of those three doctor, three dermatopathologists as ground truth. There's really 203 lesions that were enrolled in that. Sensitivity on that set was 79%, specificity 80%. If we look at tests that were run on archived samples, where again, all lesions have been biopsied, all had histology triple read as ground truth. We can see that the sensitivity in that setting for 195 consecutive, or 195 samples, sensitivity was higher, 95% specificity, 55%. If we look at it for the three gene tests, 103, you know, selecting out lesions, archival specimens, 103 samples of which 30 were melanoma, then we can see that the sensitivity there is 97%. Registry studies, by contrast, really are real-world use. There's one local pathologist whose answer is used as sort of the for calculating sensitivity and specificity and they rely on the assumption that if it's negative, the test is negative, no biopsy is done, we're going to count that lesion as benign. So greater than 11,000 lesions, so much wider, bigger data set there, but we really can't calculate sensitivity or our specificity without histology on all. And so we did see that, you know, the negative predictive value and we went back and looked at negative tests was about 1%. And we can also see that, you know, there are reader studies that can be done and that have been done. This is using all the same data that's been published, not new lesion sets. Pros and cons of the PLA. Pros, you know, the objectivity of genomics is nice relative to the subjectivity of the clinical exam. Also, you know, the fact is that if genomics are what drive cancer progression, really looking at sort of the first, you know, downstream event makes sense. There's also the potential to avoid biopsy complications, infections, scars, things like that. And it may provide information that's useful in lesions that are, you know, borderline histologically. Cons, not all tests will have sufficient quantity of genetic material to run the test. There's modest specificity, so that means that then positive predictive value, so patients with benign lesions are still going to have positive tests and still undergo a biopsy. The cost of the test is greater than that of the biopsy. And it's also not clear how prospective clinical trial data may, that probably doesn't perfectly mirror and reflect real-world use. The next application I'm going to talk about is classifying histologically equivocal biopsy and melanocytic lesions. So looking at the reader studies and data from the PLA assay tests, we know that in this, in those studies, 11% of melanoma cases were actually excluded because there was the inability to get consensus among three dermatopathologists. And the Kappa score for agreement on cases among them was 0.78, so moderate agreement, but not, it's not perfect agreement. And if we just said, let's consider on those cases where we've got triplicate read, all three expert dermatopathologists agree on the diagnosis, let's consider that to be the gold standard. And then we went back and looked at what did the local pathologist call that lesion. If we just considered, you know, what was the sensitivity and specificity of the local pathologist, it was 89% sensitivity, 91% specificity. So not perfect agreement, it's not always easy to make this diagnosis. So there are a couple of gene expression profiling tests that can be used. One is the MyPath lesion, or MyPath assay. So this looks at a series of distinguishing genes, all shown here in this table. The most distinguishing one is actually PRAME, same as what we saw in the pigmented lesion assay. And we could see in a validation cohort that the sensitivity of this profile was 90%, specificity was 91%. This is an independent validation study performed independent of the company with three different reviewers. We can see 1,400 lesions were enrolled. Once you excluded samples where there was an indeterminate score on the test, or where you couldn't get triple concordance among the panel of dermatopathologists, that number went down to 860. And then a few cases were also excluded due to insufficient material. And so what we can see here is that on this independent validation, the sensitivity was 91.5%, specificity 92.5%. And so this is pretty much an agreement with the other published studies. Now probably what really matters is correlation with clinical outcome. So you know, what we really want to know is not what will a dermatopathologist call this lesion, but is this lesion going to harm somebody, yes or no. So in this case, they looked at only those lesions that were really determined by a panel of seven dermatopathologists to be diagnostically uncertain lesions. And then they looked at the clinical outcome. So either it was a malignant outcome, they had distant metastases, or a benign outcome, no recurrence or metastases. And so what we can see here is that the lesions that were classified as likely benign, 63 out of 68 of those had benign behavior. Those that were classified as likely malignant, 47 out of 50 had malignant behavior. And of those that were indeterminate, five out of seven had benign behavior, two out of seven had more malignant behavior. So this shows that the results really did correlate, not just with what a panel of dermatopathologists said, but with actually clinical outcome. This is an additional newer test that looks at the same thing, it's just using a different panel of genes. In fact now when you, this is a test called DiffDx, when you order this test, they really are both run by the same company, so you'll get MyPath or this one if you order it. And that's a 35 gene expression profiling test. This is the one that's run second, MyPath is run first. If you get an indeterminate risk, then this one can be run because there's fewer results that come back as indeterminate with this test, the 35 gene test. Okay, pros and cons. Pros, objective data again. Cons, we really don't have a gold standard other than the lesion metastasized and caused harm or it did not. So we're really, for the most of these studies, relying on consensus of pathologists to test a test that was developed to help when pathologists have a hard time deciding whether or not a lesion is benign or malignant. Okay, so now I'm going to talk about prognostic GEP. So this is used to say, we have a lesion, we've biopsied, we've determined it's invasive melanoma, what is the likely outcome going to be for this patient? There's a few potential uses of this. One is to avoid certinal node biopsy in certain cases. The next is to guide surveillance. And the third is potentially to guide therapy. So there's three different tests available. There's a 31 gene test, Decision DX. This can integrate tumor and patient factors if you have IGEP. There's an 11 gene test that just looks at gene expression. And then there's an eight gene CPGEP test that looks at the expression of eight different genes and integrates both tumor and patient factors as well as the gene expression profile. The 31 GEP test is the one that is most commonly used in the U.S. currently and that has the most published data. This is a meta-analysis from 2020 and what we can see here is looking at the result here. So this is looking at sort of a binary or quaternary test results. So class 1 is low risk, class 2 is high risk. A and B within each is lower and higher. What we can see here is that for the lowest risk class 1A lesions, five-year recurrence free survival is 91.4 percent and for the class 2B, the highest risk patients, it's 43.6 percent. This really correlates and shows sort of this linear relationship with five-year distant metastasis free survival as well. You can also use clinical pathologic features. This will incorporate things like not just the gene expression score but the Breslow thickness, the patient age, the mitotic rate, ulceration state, and gives a percent likelihood of having things like a positive sentinel node. So these numbers can then be used to sort of match what we think about with, you know, NCCN guidelines or talk to patients and say we think that the likelihood that your sentinel node will be positive if we do it is 15.1 percent. And so this can be, you know, how could this be helpful? Probably primarily in reducing the number of sentinel node biopsies and particularly for this somewhat more equivocal cases. So if we look at T1A high risks, these are T1A lesions that for example have a positive deep margin on biopsy or T1B lesions where the guidelines would be to discuss and consider sentinel node. If you think that in most cases without any additional information, we would do a sentinel node 100% of the time. If you use, if you take that IGEP result and say it's not recommended if that number is less than 5% or you should consider it if it's 5 to 10 percent, you could see that you would be able to, you'll recommend against sentinel node biopsy for about two-thirds of T1A high risk and about 40% of T1B lesions. So this could be a way to save or reduce the number of sentinel node biopsies. You can also use this to predict, you know, sort of more distant outcomes. So recurrence-free survival, distant metastasis-free survival, and melanoma-specific survival. So this is a different percentage. It predicts each one of these. It uses both the GEP score, age, Breslau thickness, mitotic rate, and then also things like the end status, ulceration status, and tumor location. So you can see here that we can pick, you know, where would we consider a patient to be, you know, low risk versus high risk. We can pick, you know, some AJCC staging like the dichotomy between stage, you know, 2A and 2B because that's maybe where we would not versus would consider adjuvant therapy. But you can really set this wherever you want and we can see here that in blue for each one, this is where the low risk group, you know, fell versus black is the high risk group. And so we can see that the outcomes in terms of survival are very different between the two. Okay, and this was just a model that was done independently looking at, you know, how useful might this be in predicting the risk of metastasis or the risk of a positive sentinel node in patients with melanoma. And so the authors here found that, you know, compared with sentinel node biopsy for all patients, use of this model would equate to a 23% decrease in sentinel node biopsies among patients with T1B. And it also resulted in a pretty significant decrease in among those patients with T1A tumors. The next one that we can talk about is CPGEP. This is an eight gene test that uses age, Bresla thickness, and gene expression validated on a smaller set of lesions. Really, in general, thicker lesions here, not as many T1s, but we can see it as a high negative predictive value of 93.8%. And for T1, for T1 lesions specifically, and for T1 lesions could potentially reduce sentinel node biopsy by about 53%. We can see that this tends to go down as we get to thicker lesions where it is more likely to have a positive node. The authors also looked at the development cohort in this test and these are all patients with sentinel node biopsy at Mayo Clinic who did not have completion lymph node dissection and who, you know, had either a melanoma thicker than a millimeter or one that was 0.75 to 0.99 millimeters and ulcerated with mitoses or young age. And so what we can see here is that they say that, you know, by using this test and only doing sentinel node biopsies on patients who had a positive test or a high-risk test, you could reduce the complications of surgery including, about by 59%, including seroma, infection, lymphedema, and hematoma. 31GEP has also been linked to SEER database and so in this case we can look at sort of follow patients and look at their outcomes and correlate that with their GEP stage. Here we can see that patients with class 1a, the top bar, definitely have better melanoma specific survival and overall survival. Patients with class 2b, the highest risk, also tend to do most poorly. What is interesting is that the authors also just looked and said, let's look at those patients who have a GEP result and those who do not. They did propensity score matching to say here's our cohort of GEP tested patients. We're going to look at a bunch of factors and find the best kind of case control pairing for each patient who, a control for each case who did have GEP testing. So this isn't looking at what did the GEP test show, it just is looking at did they have it done, yes or no. And we can see here that you actually see better, you know, better three-year melanoma specific survival and overall survival among those patients who just had the test than those who did not. So it may suggest that that test is in some way, you know, driving how care is provided or the treatment that's used or the amount of surveillance. And then this is the 11GEP test, it's called melagenics. Patients get either a low or high score and if we look here we can see that 20 out of 127 with low score had a positive sentinel node met versus 42 out of 164 at the high score. And if we look at five-year disease-free survival, it's 95% for patients with a lower score and 78% for patients with a higher score. And this is just looking at how this breaks down melanoma specific survival by stage for stage 2 and 3. You'll notice stage 1 isn't here. There were no melanoma deaths in stage 1 patients so it didn't make sense to necessarily use this test. And there were 8 recurrences, 5 of those 8 had a high risk score. Okay, so here's some examples of using this clinically. 68 year old man referred by his PCP for a lesion of concern. You could see this is very ill-defined but it looks like it's something that definitely needs a biopsy. But you know where does that lesion begin and end? It's hard to tell. So you know so we got so you know with the poorly defined margins it's hard to go say let's go do a wide local excision. However, this patient does meet criteria for sentinel node at the head and neck. We also know that sentinel node of the head and neck can be have a low negative predictive value so more likely to have false negatives with that. So this patient did have a sentinel node biopsy. It was negative. He also had this test done. It was a high risk 2B and so based on that he has received more enhanced surveillance follow-up Q3 month imaging and he actually just recently did present with metastatic disease. This is another patient who has this very large lesion behind his ear. Depth of invasion was 0.7 millimeters at least. It was positive at margins. It did not have a high mitotic count and it also had an adjacent basal cell carcinoma that was going to be something we want to treat. And so after doing that original biopsy we also did some scouting biopsies so we can see that this really looks like in situ at the periphery and then this 0.7 millimeter melanoma. So we talked to the patient about options including, you know, staged excision alone, having a sentinel node biopsy, and then looking at test results to help guide us and he opted for the latter. So this patient had a class 1A lesion. His likelihood of sentinel node positivity was 2.9%. Remember in general we will recommend biopsy if the likelihood of sentinel node positivity is 5% or greater or 5 to 10 discuss and consider greater than 10% recommend. This is a patient who was 11 weeks pregnant, came in with this mole originally seen by e-dermatology and then brought in her biopsy and this was her result. It was a 0.9 millimeter melanoma, one mitotic rate of one. Slide sections are free. So, you know, we talked about this patient tumor-borne. It's actually recommended that you don't do sentinel node biopsy in the first trimester but that you can get excision with local anesthetic. And so we looked at her results, her class 1A result, and we can see that her likelihood of sentinel node biopsy in this case was 7%, so higher probably because she's younger. And so we did discuss and consider it and she actually opted out and to be followed clinically. It is important to mention that NCCN guidelines do not recommend for the use of GEP. Pros and cons. Pros, you can get more individualized information than you do from AGCC staging alone. This can help you to reduce the number, to reduce the need for sentinel node biopsies and may, you know, provide prognostic data that you can't get with sentinel node alone. Cons, it's not prospectively validated in randomized trials for the most part and it may not be equally informative for all stages of melanoma. So perhaps a class 2B for a 0.3 millimeter melanoma probably is different than a class 2B for a 4 millimeter melanoma that's ulcerated. That's really getting back to the pretest possibility, pretest probability of being positive. So future directions. I think it's important for us to think about which clinical scenarios is this useful for. It's probably not every single lesion that we biopsy that's an invasive melanoma. How is it going to, I think we need to think about how this might impact things like melanoma over diagnosis or over biopsy, particularly GEP. Looking at what is the role of GEP in melanoma risk stratification. We use it to help decide, you know, do you need a sentinel node biopsy? Who should get a sentinel node biopsy? How about which patients should receive systemic therapy? If they have stage 2B or 2C disease, it may, they would be eligible for adjuvant immunotherapy or a 3A. It might be helpful to have another data point. And then I think we need to agree what level of data are needed to integrate this into NCCN guidelines. Also important to note that these GEP tests, again, not under FDA approval, so it's important that we understand the data. So hopefully you understand the data a little bit more after all of that and I thank you for your attention.

Dermatologist

Laura Farris

molecular tests

melanoma

gene expression profiling

pigmented lesion assay

prognostic tests

sentinel node biopsy