Good afternoon, and welcome to Castle Biosciences DecisionDx-Melanoma webcast. As a reminder, today's webcast is being recorded. Doctors Guenther and Goldberg will begin with a presentation that follows the accompanying slide deck, followed by a brief question- and- answer session. I would now like to turn the call over to Dr. Matthew Goldberg, Senior Vice President of Medical for Castle Biosciences.
Thank you, operator, and good afternoon, everyone. My name is Matt Goldberg. I'm a board-certified dermatologist and dermatopathologist, and I serve as Senior Vice President of Medical here at Castle Biosciences. Before we get into the DECIDE publication itself, and before Dr. Guenther takes us through the data in detail, I want to spend a few minutes setting the clinical stage for why this study matters. My goal here is to ground the discussion into questions physicians are actually trying to answer in early-stage melanoma and then to level set that DecisionDx-Melanoma is already a well-validated test with a strong evidentiary foundation. This foundation includes retrospective and prospective studies along with meaningful real-world evidence.
From this perspective, the DECIDE trial is best viewed not as the beginning of the story, but as an important new perspective, multi-center addition to an already strong body of evidence supporting DecisionDx-Melanoma in Stage I through III cutaneous melanoma. For my part of the discussion, I'll attempt to do four things here. First, I'll briefly frame the key decision points in early-stage melanoma. Second, how DecisionDx-Melanoma was designed to answer two important clinical questions. Third, walk through the validation work that's already existed for the DecisionDx-Melanoma test and introduce the DECIDE study here before handing things over to Dr. Guenther, who will review the publication in more detail that was just published in Future Oncology.
With that onto the next slide, I wanted to start with the practical decisions physicians are making when they see a patient with newly diagnosed cutaneous melanoma. This is really the right place to begin because if we don't frame the clinical problem correctly, it's hard to identify the value of the test for the clinicians and patients who receive the test results. In melanoma, these decisions are very concrete. They affect surgery, surveillance intensity, and who can remain primarily in a dermatology-led clinical care pathway, and those who move into more intensive and multidisciplinary oncology pathways. Traditionally, staging and clinicopathologic factors are used to answer two central questions after a diagnosis of cutaneous melanoma. First, what is my patient's risk of sentinel node positivity? Second, what is my patient's risk of recurrence?
Those are the two questions that really drive patient management. The patient is thought to be at lower risk for poor outcomes. The treatment path is generally of lower intensity, and that usually means treatment with a wide local excision with appropriate margins, lower frequency follow-up, often primarily with dermatology and no advanced imaging. If a patient is thought to be at higher risk, the management plan becomes more intensive, and that can include wide local excision plus a sentinel lymph node biopsy, closer clinical follow-up, involvement with surgical oncology, potentially medical oncology, and in some cases, baseline and surveillance imaging. The key point here is that these are not theoretical questions. The risk of poor outcomes directly influences which patients stay on lower intensity management paths and which patients move into a more intensive staging and surveillance pathway. Next slide.
DecisionDx-Melanoma was designed to help answer both of these clinical questions using tumor biology from the primary melanoma. This is a 31-gene expression profile test used in patients with Stage I- III melanoma. It measures expression of 31 genes from the primary tumor and applies a validated algorithm to generate clinically useful risk information to inform clinician decision-making. First, the test provides a class result ranging from Class 1A, which is the lowest biologic risk, to Class 2B, which is the highest biologic risk of recurrence or metastasis within five years. It's important to highlight that this class result is independent of the clinical and pathologic information available to clinicians to improve the accuracy of risk prediction for their cutaneous melanoma. It also provides two individual outputs seen here on the right side of the slide.
One is the i31-SLNB result, which estimates the patient's individual risk of sentinel lymph node positivity. The other is the i31-ROR result, which estimates the patient's individual risk of recurrence. This is another important point to emphasize clearly. Predicting sentinel node positivity and predicting recurrence are related, but they're not the same clinical question. Sentinel lymph node biopsy is about nodal or the lymphatic spread of the melanoma to the draining lymph node basin. Recurrence risk is broader. It's asking about the risk of relapse more generally, whether that ultimately presents regionally or distantly, as well as including melanoma-specific survival. It's entirely appropriate and really what we would expect that the clinical pathologic inputs used for the i31-SLNB and i31-ROR are not identical.
For the i31-SLNB result, the model incorporates the gene expression profile score along with ulceration, Breslow thickness, age, and mitotic rate. For the i31-ROR, again, it uses the gene expression profile score and incorporates ulceration, age, Breslow thickness, mitotic rate, sentinel node status, as well as tumor location. This is really a strength of the test. It's really what clinically intelligent models should do here. The model should bring in the specific independent prognostic factors, clinical, pathologic, and gene expression profiling that provide the most clinically impactful results to refine risk prediction and inform specific clinical management decisions for patients with cutaneous melanoma. As it relates to the i31-SLNB and as shown on this slide, the 31-gene expression profile score was the most significant variable in predicting sentinel lymph node positivity and is an independent and significant variable in recurrence outcomes as well.
The take-home message here is that DecisionDx-Melanoma was built to answer the actual questions physicians and patients face after a diagnosis of cutaneous melanoma with outputs aligned to the specific clinical endpoint being predicted. On the next slide here, it's important to level set that DecisionDx-Melanoma is already a well-validated gene expression profile test. What you're seeing on this slide is a representation of the broader evidentiary foundation for DecisionDx-Melanoma. The studies shown here span both retrospective and prospective validation cohorts, and they consistently show that DecisionDx-Melanoma provides meaningful risk stratification across survival-related endpoints. The pattern is consistent. Patients with lower risk results, specifically Class 1A results, have markedly better outcomes than patients with a high-risk class result, specifically here in dark red, the Class 2B result. This robust separation has been demonstrated repeatedly across independent study cohorts.
When we think about the DECIDE trial, it fits this overall evidence foundation. It's important to say plainly that it builds on a strong body of evidence that exists before the DECIDE trial even begins. On the next slide here. It's important also to highlight that the evidence does not stop with just retrospective and prospective validation studies, but also includes substantial real-world data from prospective retested patients who received the DecisionDx-Melanoma test as part of their ongoing clinical care. This is what's on this slide here, the Castle's collaboration with the NCI and SEER Cancer Registry. That's really a key part of this story.
Because the DecisionDx-Melanoma test is already being used in clinical care for patients with Stage I-III cutaneous melanoma, this collaboration makes it possible to examine outcomes in a large real-world population of patients who actually received the test as part of their melanoma management. This is really what makes this analysis so meaningful. It reflects real patients, real ordering behavior and outcomes after the diagnosis of cutaneous melanoma in clinical practice here in the United States. First, in this large real-world cohort of clinically tested Stage I-III melanoma patients, the 31-gene expression profile test effectively stratified mortality risk within AJCC sub-stage groups. Second, the 31-GEP test demonstrated significant prognostic value for melanoma-specific survival beyond traditional clinicopathologic factors, providing independent prognostic information from the information that was known already at the time of diagnosis.
Third, the 31-GEP test, DecisionDx-Melanoma test, is the only GEP test in melanoma that's shown to be associated with improved survival, with clinically tested patients showing a 32% lower three-year melanoma-specific mortality rate than untested patients in this study. Again, before we dive into the DECIDE data here with Dr. Guenther, it's important to level set that the DecisionDx-Melanoma has substantial evidentiary basis behind it. This slide highlights that the story extends beyond validation studies and into real-world patient outcome data derived from clinically ordered tests and used as part of patients' ongoing melanoma care. Again, these are patients for whom testing was integrated into their actual clinical management after diagnosis. Then the DECIDE study, in this sense, adds more prospective, multi-center evidence to this already strong foundation. On to the next slide here.
As we transition into the DECIDE study, it's important to ground the discussion in how sentinel lymph node biopsy decisions and the sentinel lymph node biopsy procedure itself is used in clinical practice. In routine care, physicians start with clinicopathologic factors to estimate a patient's risk of sentinel node positivity, and the key threshold here that matters is the 5%. Current guidelines support avoiding the sentinel lymph node biopsy procedure when the expected risk of node positivity is predicted to be less than 5%. However, once a patient moves above that level, the conversation changes. In the 5%-10% range, sentinel lymph node biopsy is to be discussed and considered. Above 10%, this is the group in whom sentinel lymph node biopsy is discussed and offered, as seen here in the table on the left.
I really wanna highlight that this 5% threshold is the critical clinical decision-making line. For an advanced molecular test to be clinically useful in the setting of sentinel lymph node biopsy decision-making, it has to do more than add general prognostic information. It has to refine risk in a way that can actually guide clinical decision-making by identifying patients whose predicted risk falls below 5%, where clinicians may feel more comfortable deferring sentinel lymph node biopsy. Similarly, it has to move patients whose risk is above 10%, where sentinel lymph node biopsy should clearly be part of the conversation. This is really the evidentiary bar and why the DECIDE trial matters so much. It's evaluating whether the i31-SLNB result can prospectively identify patients who truly fall on either side of these established clinically meaningful thresholds of 5% and 10%, respectively.
On to the DECIDE study here on this next slide. This is again a prospective multi-center clinical utility study designed around a very practical question. Can the i31-SLNB result prospectively inform sentinel lymph node biopsy decisions in a clinically meaningful way? The study had three primary objectives. First, to prospectively confirm the performance of the i31-SLNB result in predicting sentinel lymph node positivity. Second, to evaluate real-world use of the i31-SLNB result in guiding sentinel lymph node biopsy decisions. Third, to assess recurrence outcomes among patients with less than 5% predicted risk of sentinel node positivity, including patients who did not undergo the sentinel lymph node biopsy procedure.
The study enrolled patients diagnosed within two months who were being considered for sentinel lymph node biopsy and whose clinicians were ordering the 31-GEP to help guide this specific clinical decision. Testing was performed, results were returned, and the patient and physician made the sentinel lymph node biopsy decision in the model of shared decision-making in a real-world clinical setting, evaluating both the class result and the i31-SLNB result. Clinical outcomes were then tracked, including sentinel node positivity rates and recurrence-related outcomes in the low-risk, including in the low-risk patients who did not undergo a sentinel lymph node biopsy procedure. This is the way to really frame the DECIDE data that Dr. Guenther will walk us through here shortly. It's not trying to collapse every biologic question into just one endpoint.
It prospectively evaluates the output that was built for nodal risk, namely the i31-SLNB, against the nodal endpoint that matters of sentinel node positivity. Then it follows recurrence outcomes in the patients identified as low risk and managed without the sentinel lymph node biopsy. In my view, the DECIDE study here should be seen as a prospective study that adds both clinical validity and clinical utility evidence to an already well-validated DecisionDx-Melanoma test, specifically around one of the most consequential decisions in early-stage melanoma management of the sentinel lymph node biopsy. Really, at this point, it's my pleasure to turn the conversation over here to Dr. Michael Guenther. Dr. Guenther is a board-certified surgical oncologist with more than three decades of experience in the multidisciplinary surgical management of cancer, including melanoma, breast cancer, and other solid tumors.
He earned his medical degree from the University of Michigan Medical School, completed general surgery residency at the University of Cincinnati Medical Center, and then completed advanced surgical oncology fellowship training at the John Wayne Cancer Institute, affiliated with UCLA. He is certified by the American Board of Surgery. He has served in high-volume oncology practices, most recently with St. Elizabeth Physicians in Edgewood, Kentucky, where he applies evidence-based surgical approaches to cancer diagnosis, staging, and therapy. Importantly, he has also been a lead author and investigator on multiple clinical studies, including as lead author on the multicenter DECIDE trial that he'll discuss here shortly. Dr. Guenther, thank you again for being here with us today, and I'll hand the microphone over to you.
Thank you, Dr. Goldberg. And again, I appreciate the opportunity to do this. Slide 13, please. I had the privilege of presenting the first iteration of this DECIDE study back in 2024 at the Society of Surgical Oncology annual meeting. We had 322 patients that followed this algorithm, which essentially, as Dr. Goldberg mentioned, began with a diagnosis of melanoma within the last two months, a patient over 18 years of age considering a sentinel node biopsy in whom we might use this 31-GEP to guide a sentinel lymph node biopsy decision. At the second visit, we ordered the test, and then we decided whether or not to perform a node biopsy. That was a mutual decision. There were times when I made a recommendation.
There were times when the patient had a strong feeling, and ultimately, we honored what the patient wished to have performed. We then either did or did not do a sentinel node biopsy with a tumor excision and then tracked the clinical outcomes of sentinel lymph node positivity, recurrence-free survival, distant metastasis-free survival, and melanoma-specific outcomes in those with a less than 5% risk of nodal metastasis. Next slide, please. The demographics of this study. There were 912 patients. We had an average or median age of 65 years of age between 20 and 90, so we had a nice broad group. The Breslow's depth median was 0.8 mm and ranged from 0.1 mm- 12 mm. This reflects typical suburban dermatology practices. Interestingly enough, 42% of the patients had a transected base.
This also reflects the pattern of shave biopsies in clinical practice and is quite reassuring to clinicians that the assay still applies and functions properly even in that setting. Overall, 47% of the patients had a sentinel node biopsy, of which 10.2% were positive. For those not undergoing a sentinel node biopsy, which was 52.9% of the group, nearly 90% were either T1a, T1b, or T2a primaries. We did have 10% of the group that had deeper lesions, which, reassuringly did not have node biopsies performed in some instances. Ulceration was present in 9% of the cases. Mitosis was not a frequent finding. Overall, we had a 52% yield of less than 5% nodal predicted risk.
In that discuss and consider group from 5%-10%, it was 32.2%, and 15.8% of the patients had a predicted node positivity rate of greater than 10%. Slide 15, please. In this subset of Stage IB patients, which consists of T1b and T2a lesions, for the non-clinicians in the room, a T1b lesion is a lesion less than 0.8 mm with ulceration or high-risk features, and a T2a lesion is a lesion between 0.8 and 1 mm with or without ulceration. Stage IB, the population of most concern, the sentinel node positivity rate, if you had a low predicted node positivity, was 1.4%. In contrast, if you had a high-risk GEP prediction, the sentinel node positivity rate was 18.5%.
That's a 13.2 times magnified risk between low risk and high risk. That is a substantial amount of stratification within the same stage, and that is very valid for clinicians. We reached our threshold well below 5% in the low-risk lesions and had a much higher yield in high-risk lesions. Next slide, please. Again, in looking at this Stage IB, the most interesting and, if we say, controversial group, we had that 1.4% yield for low-risk lesions and an 18.5% risk for high-risk lesions from the DecisionDx i31-SLNB. Last year, the MERLIN_ 001 study was presented and unfortunately had a 6.5% node positivity rate in the low-risk group, and the high-risk group was 2.8 times higher.
It was 18.3. The low risk i31-SLNB was more than 4 times lower than the CP-GEP assay was, while the assay only separated risk 2.8 times higher in the CP-GEP group. We're well below that 5% threshold with the one assay, but not so with the other, and high-risk lesions were 13 times more likely to have a positive node. Next slide, please. When broken down by the entire group, a T1- T4 primary all-comers, if you had a less than 5% prediction, the yield was only 2.6%. In the 5%-10% range, the discuss and consider range, it was 7%, and if it was greater than 10%, it was 21.4%. That's an 8-fold magnifier between less than 5 and greater than 10.
On the other hand, in the group of most concern, T1b and T2a, only a 1.4% node positivity rate, which then went from 7.4% - 18.5%, a 13-fold magnifier as the risk went up. We had an extremely successful assay at predicting who had low risks of node positivity. On the other hand, every 5%-10% is a discuss and consider, and most of those patients were going to get recommended to undergo a node biopsy, and especially with a greater than 10% risk. Next slide, please. Another way of displaying this, the AJCC staging never gives us in this T1b or T2a group, Stage IB, a circumstance where we avoid sentinel node biopsy.
The Merlin GEP assay, again, never reaches the threshold of 5% and then also demonstrates a high-risk lesion. The DECIDE study has confirmed for us a very successful prediction of low-risk node positivity, as well as the 5%-10% discuss and the greater than 10%. This stratification is very important because we can precisely identify those high-risk patients who need node biopsies, as well as those truly low-risk patients who could safely consider foregoing a node biopsy rather than being in the discuss and consider range. Next slide, please. This can't be emphasized enough. This discrimination in the same stage between a low-risk lesion and a high-risk lesion is like wiping fog off the windshield. We can see that 13-fold difference between low risk and high risk. This is clinically very actionable, and this is an important assay quality. Next slide, please.
The NCCN guideline of 5% is important to discuss. We typically do not perform sentinel node biopsies for lesions under 0.8 mm without high-risk features because a very important clinical study called MSLT-I was performed in the 1990s. It randomized patients to a wide local excision or a wide local excision and sentinel node biopsy. A lymphoscintigram was performed pre-operatively. The surgeon removed a node or nodes from the basin or basins at risk. Those were read as clean by the pathologist, and yet the node failed in that basin at a 1% per year risk, which came up to be 5%. It made no sense to look for something with a less than 5% incidence when the error rate of the test was 5%.
In that setting, a pathology report on a 0.6-mm or 0.7-mm primary with no adverse features would estimate a one in 20 chance of a node being positive. If no nodes were biopsied at all, one in 20 patients would have a recurrence. That was set up as the guideline by the NCCN. Last year, Dr. Prieto presented an accumulation of studies. The CP-GEPs for T1 and T2 lesions had a false negative rate of 6.2% and a true negative to false negative rate of 15: 1. That did not yield even equivalence with the NCCN guidelines. The GEP or the i31-SLNB study from his pooled results showed a 2.8% false negative rate and a 34: 1 true negative to false negative ratio, again, exceeding the NCCN guidelines.
I'm happy to report that the current DECIDE data was only 2.6% false negative rate for all comers, and again, a 37:1 true negative to false negative ratio. For the group of most significance, T1 and T2a, it was only a 1.8% false negative rate and a 55:1 true negative to false negative ratio. This was very specific and outperformed not only the NCCN guidelines, but substantially outperformed the CP-GEP test, which did not appear to function as well as a simple pathology report. Next slide, please. What happened to those patients with a low-risk lesion who did not have a sentinel node biopsy? They had a 97.8% 3-year recurrence-free survival. They did extremely well.
If they had a high-risk lesion, they had a nearly 7% difference in recurrence-free survival at three years. This was one of the main pillars of the study to demonstrate that low-risk patients could forego a node biopsy without jeopardizing their risks, and they were, in this study, able to do so. Next slide, please. The identification of high-risk patients in a traditionally, quote-unquote, low-risk cohort is incredibly important and is one of my real passions. There is a false consensus or belief that shallow lesions rarely, if ever, spread. That is not true. Next slide, please. For a simple T1a lesion, last year in a large study in Europe, the CP-GEP assay did not identify a single high-risk patient among 714 patients.
In contrast, a large JAAD study performed last year using DecisionDx found an 8% yield of non-1A lesions, Class 1B, 2A, or 2B lesions. That was an 8% cohort within there that had almost 6% higher risk of death. As a clinician, these are patients who are incredibly important to find because they have shallow disease with bad biology. Next slide, please. An example of this is a patient of mine. This is a 45-year-old man who presented with a 0.4 mm lesion on his back. Ulceration was present, as was regression. No mitosis. This is Stage IB. If you ask a group of dermatologists or primary care physicians what they would do with this, most of them would do a narrow excision in the office, and that would be it. If you follow NCCN guidelines.
This is Stage IB, It's in the 5%-10% yield, so we should discuss and consider sentinel node biopsy. I did not do that. I ordered a DecisionDx, which had a Class 2B score, and the gene expression was 0.91, one of the highest I've ever seen. It had a risk of recurrence at five years if the nodes were clean of 88.8%, a node positivity rate of 19%, and if it were Stage III, the recurrence-free survival drops by another 10%- 78%. In that setting, this person underwent a sentinel node biopsy, which demonstrated one out of six lymph nodes with 'melanophages.' It was later interpreted as negative at an outside hospital. No further treatment was rendered, and he returned a year later with a lemon-sized mass in his left axilla, which was metastatic melanoma. The CAT scan shows that.
He underwent two cycles of immunotherapy and a salvage dissection, which demonstrated one large positive lymph node. This is a shallow lesion that most clinicians, I dare say, in the absence of gene expression testing that is reliable, would have simply performed a narrow excision and would have missed his life-threatening Stage III disease. This is a really important part of this assay's utility. Next slide, please. Dr. Goldberg touched on recurrence, which is where this assay has its particular strengths. If we look at the distribution of gene expressions from 0- 1, where 0 is our lowest risk group and 1 is the highest, thankfully, 2/3 of these were in the lowest risk of metastasis. 96% are recurrence-free survival at 5 years. These are people for whom de-escalation of therapy and follow-up should be considered.
They need to follow up with dermatology for new primaries, but they don't need expensive testing, imaging, and anxiety-provoking high-risk visits. They need to follow with dermatology. On the other hand, the higher-risk lesions had a 61% 5-year relapse-free survival. Nearly 40% of them recurred, and there is clear evidence that those patients benefit from enhanced surveillance and potentially even life-saving treatment such as surgery or immunotherapy. This ability to separate 35% within the spectrum of outcomes is a significant factor in the management of patients and is not present in other gene expression assays. Next slide, please. We have strong prospective data now, and there is more coming as the future goes by, that we have accurate risk stratification. Low-risk test results are associated with extremely low sentinel node positivity outcomes.
This i31-SLNB gives us great confidence in identifying which patients can avoid sentinel node biopsy. Should they choose to do so, we had a 2.2% recurrence rate at three years. The assay performed the way we had hoped for, and it validated the initial DECIDE study. With that, I think we'd like to thank the folks for listening to this and be happy to take some questions.
We'll now begin the question- and- answer session. If you would like to ask a question, please press star followed by one on your telephone keypad. If, for any reason at all, you would like to remove that question, please press star followed by two. Again, to ask a question, please press star one. Our first question comes from Catherine Schulte with Baird. You may proceed.
Hi. Thanks for the questions, and thanks for this run-through on the data. I wanna just be clear on kind of the incremental takeaways from this data versus what has previously been published. I mean, you guys have put up strong data on the basis of meeting this kind of NCCN 5% threshold on sentinel lymph node biopsy guidance decisions. What is kind of the key incremental from this data? You know, now that you have this prospective multicenter data showing a strong, you know, sentinel lymph node biopsy prediction, what is kind of that missing link for NCCN? You know, do you think this directly addresses some of the level of evidence concerns that they have raised in prior updates?
Catherine, this is Matt Goldberg here.
Oh, go ahead, Matt.
No, go ahead, Dr. Guenther.
I was gonna say that the performance of the assay has never been challenged. The reliability, the validation studies, the enhanced numbers through DECIDE have all pointed at the same thing. There has never been a contrary publication that has questioned the validity of either the assay itself or the interpretations of this. The takeaway from the surgeon's point of view is that we need to do less negative sentinel node biopsies, and this is a huge step in that direction. It is also a conviction that we need to do evaluation of shallow lesions that have bad biology because it affects patient outcomes.
The perhaps unsung value to this from the clinician's point of view is the recurrence data, which is truly where we affect patient outcomes. We have the ability to become efficient, giving patients the treatments they need, but avoiding the treatments and tests that they don't. As far as the NCCN guidelines go, I don't think that this data or that the outcomes from this in any way were challenged by the NCCN. I think that was a consideration for a very narrow subset of patients who would challenge or take a less than 10% yield of a node biopsy in return for some uncertainty. It doesn't affect the 215,000 assays that have been performed or the 13,000 physicians that use this reliably. That data is as strong as it ever has been, and I believe DECIDE just emphasizes that.
Yeah. Catherine, I'll just echo Dr. Guenther here that the incremental value here for the DECIDE trial, I think to Dr. Guenther's point, is the consistency from the retrospective data that's been published from manuscripts like Whitman et al. on the i31-SLNB's performance and now in this prospective multicenter cohort showing again, consistent accuracy of the i31-SLNB, the ability to identify patients below the 5% threshold at that clinically actionable threshold that remains the threshold in clinical practice and identifying patients at high risk, so over 10%. It's that delta between low risk and high risk that the DECIDE trial highlights with the consistency from retrospective to this prospective multicenter data here from DECIDE.
All right. Perfect. Thank you.
Thank you. The next question comes from Thomas Flaten with Lake Street. You may proceed.
Yeah. Hi, Dr. Guenther, a couple of questions for you. Did you have patients in your enrollees who had a sentinel lymph node biopsy that turned out to be positive? What was their three-year performance on a recurrence-free survival basis?
I most certainly did. I think I've had 145 patients I've put on this trial, so I most certainly did have patients that were positive. I can't tell you my exact numbers on my own Stage III, but I can tell you what my high-risk lesions, what happened to them. I published that, and we had 30% and 40% recurrence rates on our node-positive Class 2B lesions. They are a completely different subset than any other melanoma population. Node-positive high-risk lesions do very poorly. Node-positive but low-risk lesions with good survivals, they do quite well. My data reflects almost word for word what the Castle database shows.
Got it. Then in kind of real-world clinical practice, when you sit down with a patient who might be in the less than 5% AJCC category and you present them with a Class 1 result from the Castle test, what goes into that? Like, can you describe some of those conversations? I'm just curious to hear what tends to sway opinion, what tends to push some patients towards getting a biopsy regardless of what data you show them, just some of the real-world practice implications of bringing this data to bear with the patient.
What a great question. First off, just as a background, melanoma is a scary word, and most of the patients that I see are convinced that they are going to have a terrible problem. When I bring up a discussion of what we could do with this, that we could essentially predict the future on their outcome, I've never had a patient say no. The ones with low-risk lesions were overjoyed they didn't have to see me again. The high-risk lesions, they always thought they were in trouble anyway, so they were reassured that we took it seriously. Having said that, our risk numbers tell us that we start to see lesions that spread at 0.3 mm. Oftentimes under 0.3 mm, we are not aggressive there.
However, if this is your partner's wife or this is a nursing student or this is a transplant patient or someone who's incredibly anxious, the value of giving good news to a patient is definitely underrated. I give so much bad news as a surgical oncologist that giving good news to a patient that they have a survival of a Stage I lesion that would be 99.8% or even a Stage II lesion that was in the high 90s is invaluable. It's life-altering. The discussion of how would you like to know what this lesion is likely to do, almost every person says yes because it affects the outcomes of their lives so much. Like I said, there are exceptions to the low-risk rule.
I typically don't think of this for people under 0.3 mm unless there is an extenuating circumstance, and anxiety relief is one of those. You know, the husband or wife of three kids who's 30 years old with a lesion that comes back as a Class 1A with a recurrence, you know, of 90%, 97% or 98% and a node positivity rate of, you know, 1% or 2% is palpable. That discussion is real world in the office, and you put your finger right on one of the main uses.
Just final question. You emphasized the point on one of the slides that in the T1a patients, the competitive test didn't discriminate any patients into a higher risk category, whereas DecisionDx-Melanoma did. I think it was 8% were something other than Class 1A. Did you have that experience in the DECIDE study where you had patients that were up-risked?
Absolutely. It's a critical performance for this. Overall, the Castle data under 1 mm is 15% have a metastasizing lesion. Yes, I did, and that's why I order this in that subgroup. It is sometimes like pulling teeth to get people to think of that. I've had dermatologists argue with me that shallow lesions don't spread. If you look at the survival, 63% of the deaths from melanoma were in patients who were Stage I or Stage II at diagnosis. Their nodes were clean. That is a big numerator. I know the denominator is big as well, but a lot of deaths, dare I say the majority of deaths, come from people with unrecognized biologically aggressive disease. The death is not the only factor that goes into that biological aggressiveness.
Great, I appreciate that. Thank you.
Thank you. As a quick reminder, if you'd like to ask a question, please press star one on your telephone keypad. The next question comes from Puneet Souda with Leerink Partners. You may proceed.
Hi, everyone. This is Carlos on for Puneet. Congrats on this publication. I've got two questions. The first one, I know it's still pretty early, just, you know, about 1.5 weeks , but if you could give any color on how the sales force is utilizing this study and what reception from the doctors that you're speaking with has been so far, that would be really helpful. Thank you.
Carlos, did you want to direct that maybe at Derek or Dr. Goldberg?
Yeah, I think that's primarily at Derek. Thank you.
Okay. Can you restate that question one more time, Carlos?
Oh, just on how you're utilizing this study so far, and what reception the sales force is hearing from doctors?
Sure. Anecdotal right now, just so we're clear on that fact. There are sort of three elements that this study overall communicates. Not only this paper here, but also the earlier report. One of them is to say, your peers are using our tests, I mean, last year, nearly 40,000 times, to help make early stage early decisions in the treatment of these recently diagnosed patients. What we find is that just behind the patient's decision to undergo or not undergo this procedure, the very next thing is our test. That tells you, I think, that clinicians who use the DecisionDx-Melanoma test to make sentinel biopsy guidance decisions actually are using the test from a clinical practice standpoint.
That's a great encouragement to say, "I'm not an outlier. I'm actually part of adopting good patient care advances with their technologies." The second element is, as Dr. Guenther went through, does your test consistently identify people who it claims are at a low risk of having a positive sentinel lymph node, are they actually below that 5% standard that's been around for, I would say, several decades, Dr. Guenther, probably since the late 90s, maybe early 90s. If that's the case, then I have faith that I should be using the test to help make that decision. It reinforces earlier data showing that time and again, our test has been able to demonstrate that if we say you have a likelihood of being less than 5%, you actually are below 5%.
The third data element is what happens to these patients if the physician and the patient actually follow a test result and say, "You're predicted to have a likelihood of sentinel lymph node biopsy positivity less than 5%. We're gonna have a discussion here and avoid that procedure." How do they do? Are we potentially doing harm by not following through with the sentinel lymph node biopsy surgical procedure? The answer is no. These patients do extremely well in terms of a low likelihood of having a metastatic event, which is a great way to wrap around good patient care going forward.
Those are the three elements that this study, I think, lets us communicate well to clinicians. Now, the feedback has been, I think, very similar to Dr. Guenther's commentary on the call so far, which is to say, the data is pretty straightforward. It's not unique in that it replicates all prior studies, both in terms of predicting sentinel to positivity, plus also demonstration of a low risk, you know, a Class 1 test result or a low chance of recurring, actually really low in practice. It reinforces those things. As Dr. Guenther talked about with his case presentation from his practice there, we're also trying to find patients that look like they have, I wouldn't say nice shallow, I think was the word Dr. Guenther used, shallow melanomas that are harboring really aggressive biology, and we've shown that as well. Those are the elements that we see.
The feedback is pretty consistent in that this is one more piece of data that makes me comfortable that we're doing the right thing by incorporating your test in our clinical practice. We hear that consistently up and down the chain from surgical oncologists like Dr. Guenther to nurse practitioners and PAs and dermatologists in the dermatological community. That's a great element of support to reinforce. I saw your data before. It's always good to have consistent confirmatory data, and this is just one more study which goes ahead and shows it.
All right, thanks. That's really helpful. I have one more follow-up for you. You know, I thought one of the more interesting pieces of this data was how you did have head-to-head data versus a competitor. If you could just give an update on that competitive environment and how significant this new study is as another tool in your toolbox to compete against that competitor, that would be really wonderful. Thank you.
There have, I guess, in the past been a couple of competitors. There was a U.K. firm that launched a test a couple of years ago. I think they went bankrupt maybe last year, last fall. The test, as [crosstalk], is no longer on the marketplace in the U.S. There is another potential competitor in Germany called NeraCare. They've made attempts to commercialize in the U.S., but have never done that I can see. Then we have SkylineDx, which is what you're referring to here. They launched their test, I think their press release was back in May of 2020, if I recall correctly. Second quarter 2020 during COVID. I would say looking at medical claim data.
We think they've been able to achieve, what, less than 1%, probably less than 1% market share in the U.S. marketplace over the last, I guess it'd be five years at most, 4.5 years or so. Most clinicians that we have a chance to talk to, I think, challenge us by saying, "What's the data look like? Is it consistent? Are your quote claims actually what you actually have showed in clinical studies?" We can say, "Yes, they are." I think that's a pretty common expectation there. I think that's the commentary I guess I could offer you.
Thank you.
Thank you. The following comes from Mark Massaro with BTIG. You may proceed.
Thank you. This is Vidyun on for Mark. Thanks for taking the questions. I actually just had one for Dr. Guenther. Just wondering what value does DecisionDx-Melanoma add in those borderline cases that you outlined, considering sentinel lymph node biopsy when risk is between 5% and 10%? Just what does the treatment pathway look like for those cases that are more borderline? Thanks.
Another excellent question. I think one of the shortcomings would be people who only look at the node positivity rate. The node positivity rate comes on every test, and it's personalized and it's precise, but it has to be taken in the context of what is their recurrence-free survival if the node is negative or if it's positive. For instance, a 5.2% yield of node positivity, when the recurrence-free survival is 95% and 94%, respectively, wouldn't make a very big difference, especially if it was on a head and neck area, in a cosmetic area on a young woman, for instance. All of those need to be factored in. You can't look at just the node positivity. If someone is not a surgical candidate, I typically don't order the test.
If they're on home oxygen at the end of life, then the test, I wouldn't order it because I wouldn't be anticipating a node biopsy. It's all the data and the clinical factors and their own emotional status and family circumstance all get factored into that 5%-10%. Thankfully, over half the people in this study got clarified down into the less than 5% risk, but you can see our node positivity rate was over 7% in that 5%-10% range. Those people who, if they have a positive node, are gonna be candidates for either close follow-up or immunotherapy. If they have a negative node but a high recurrence rate, they're also gonna get rolled into close follow-up in a survivorship program.
The way that information is actionable is if the node is negative and they have low risks, then we typically would talk about potentially avoiding it. In our data hands, 7% is not 5%, it's 7%. If those people have measurably high recurrences and the node would make a difference, then we oftentimes do offer that to them, especially if it's not gonna be a sensitive area or cosmetically disfiguring. That 5%-10% area is, as they say, discuss and consider, but most of the time I recommend considering it because of the therapies available and the close follow-up that we can do, whether the node is positive or negative.
Just to echo, to add just slightly here on Dr. Guenther is just how critically important it is. You hear how Dr. Guenther framed the response to have confidence in not only the point prediction of, you know, sentinel node positivity, but also have confidence in risk of recurrence information. That's part and parcel of clinical decision-making at the bedside. The data presented here from Beaird et al. from the DECIDE study really demonstrates high quality of both endpoints, not only of the point prediction for sentinel node positivity, but also robust risk of recurrence information, the Class 1A result having that outstanding performance in the risk of recurrence data that was just presented as well. It's critical to have visibility to both endpoints when having clear conversations with patients at the bedside.
Perfect. Thanks so much for the time.
Thank you. As a final reminder, if you'd like to ask a question, please press star one. The next question comes from Mason Carrico with Stephens. You may proceed.
Hey, guys. Appreciate you guys hosting the call here. I guess what do you see as the primary reason DecisionDx-Melanoma hasn't been included in NCCN guidelines despite, you know, multiple published studies showing its ability to identify patients below that 5% threshold? I mean, is it study design, patient sample size, the duration of follow-up, I guess? I'm just trying to get an understanding of what you guys view or what they believe or what they're pushing back on just given the, you know, large body of clinical evidence that you've published around this test.
Sure, Mason. Derek here. Camilla?
No, that's what I was gonna say. We'll hand that one over to Derek, yeah.
That's an excellent question there. One can look at our data and the consistency over 53-55 publications, a couple of which are the focus of systematic reviews and meta-analysis, which is the highest level of evidence, of course, and ask why not? That's a very excellent question that we would hope when we look at this last publication here from Dr. Beaird and colleagues, that they'll look at this data and say, "You've got a whole swath of prospective studies, of retrospective studies, of clinical use studies. Of studies like this one here showing that if you avoid a sentinel biopsy procedure, because of a low-risk prediction, patients do extremely well." It's hard to imagine why this would not be in the guidelines in short order. I'd agree with you on that one.
Got it. Maybe a similar question. Given the data that you published, the utility you've shown in this test, what objections or misconceptions maybe, if any, do you encounter from dermatologists who are still hesitant to use GEP testing to guide either sentinel lymph node biopsy decisions or, you know, incorporate the risk of recurrence?
What are the kind of questions or pushback that we hear?
Yeah. Like I said, similar to the last question, the evidence that you published, when you talk to dermatologists, what is the common objection or even misconception that they kind of present, the dermatologists specifically are who are hesitant or choose not to use this type of testing?
There's probably a couple of buckets there that I think have pretty clear delineation, and there's probably some squishy areas. The clear delineation buckets is kind of echoing Dr. Guenther's earlier comment that when you hear the word melanoma, it always must be bad. There are a number of dermatological clinicians who, if they biopsy a suspicious-looking mole that comes back invasive melanoma, that patient is going to a surgical oncologist, a medical oncologist for follow-up and care. It's basically kind of a patient being exited out of their practice. Those dermatologists have a low level of comfort of treating that form of skin cancer.
To them, ordering our tests when they're sort of graduating their patient to a different care pattern, "Why am I ordering your test?" I don't know how big that population is, but that was bigger a few years ago. It's less big now, I think. That's clearly one group of dermatological clinicians. I think the other group is to say, "I'm a dermatologist in the you know Kentucky, Cincinnati area where Dr. Guenther lives. He's a very communicative surgical oncologist. We have a great referral pattern back and forth. I'll let Dr. Guenther make the decision if he wants to order the Castle test when I refer an invasive melanoma to him for a consultation.
Therefore, let him make that call. That represents a certain number of dermatologists, maybe more than the first group, I guess I would call that, Mason. Then I think it's interesting, the number of new ordering clinicians that we have come on board every year has been amazing to me, I guess you would say. Some of that represents dermatologists who, even though the test has been around a while, maybe they have a sort of no see industry people policy. Maybe they're just hard to get to because they just have a very busy practice, and they're just under-informed or unaware.
The ability of our sales force, of our marketing materials to wrap around them and say, "Hey, I know you're busy, but can we sit down and talk to you about advances in the personalized management of your melanoma patients?" They raise their hand and say, "I have no idea what you're talking about." There is still a large pocket of clinicians like that where you would think it's been out there a while, what gives? What gives is that they don't wake up in the morning asking, "What does Castle have for me today that's new?" They don't even know about us and sometimes. That portion of dermatologic community is larger than you might expect, but that's also why we see the new ordering clinician growth every year.
Sometimes it's somebody who's been formed for a while, and they've seen maybe Dr. Guenther using the test clinically, and they begin to order it in their own patients in their practice. Other times they just haven't. They've just had such a low awareness of the value of our test and what will I do differently as a doctor? It just takes time to overcome that. I think those are probably the three or four main bucket groups.
Mason, could I just add one thing?
Got it. Thank you.
I get very little pushback from dermatologists. They're generally nice people. They see the data, and they say, "Yes, this is my reality. But I'm afraid that if I order this and send it to Dr. University Surgeon, that he's going to tell me or she's going to tell me, 'No, I don't use this.'" They're afraid sometimes to send this with their patients. Once they get to the point where they understand that this helps them and that they're in their patient's best interest to have this information, and that a dermatopathologist at some university is not better than a gene expression test, then we break through.
That's been my biggest frustration from dermatologists, is that they don't get surgeon buy-in because some of them are at universities where they're dominated by different rules or nomograms or things like that. That, to me, is a far bigger issue than the dermatologists who are almost always believing in data when they see it.
Got it. Super helpful. Thank you, guys.
Thank you. I will now pass it back over to Dr. Goldberg for closing remarks.
Thank you, operator. I just wanted to say thank you to Dr. Guenther for joining us here on the call and for those also who joined us on the call and for participating in this Castle Biosciences DecisionDx-Melanoma webinar. Thank you for your attention and interest.
This concludes today's conference call. Thank you for your participation. You may now disconnect your line.