Morning. My name is Derek Maetzold. I'm the CEO of Castle Biosciences. With me today is Frank Stokes, our CFO. I want to thank Jefferies Conference for the opportunity to come and present today, and look forward to taking questions when we get done. Let's talk a little bit about Castle Biosciences. First, I encourage everybody to read through our disclaimers on the corporate presentation and look at it on our website. We try and focus on improving health of patients through innovative test services. We develop those internally or look to acquire them with the key goal of saying, will this novel test have a positive impact on patient health or not? If it does, then that's something we should pursue, either investigating, discovering, developing, or acquiring. We play in a couple of locations here.
One way to think about organizing diagnostic test services is sort of where do you perhaps come and impact the patient journey. At this point in time, our earlier tests are located in sort of the diagnostic support area, which from a commercialized test right now, we have MyPath Melanoma, which really assists dermatopathologists and dermatologists in understanding if this biopsy for a melanoma is actually a melanoma or not. We then move into patients who have the definitive diagnosis on pathology typically and provide tests that we would call risk stratification or prognostic tests. That is important from our perspective because, as far as I'm aware, every disease state, once it's diagnosed, the next step is really, what are we facing downstream? Do we have a good outcome, be it in this case, usually cancer care, progression, or it could be LDL cholesterol leading to cardiovascular events or not?
You adjust your treatment pathways based upon risk stratification. One goes downstream a bit more into therapy selection process. We have one test on the marketplace, our DecisionDx-SCC test, which is for use in patients with squamous cell carcinoma of the skin, but high risk. They have the presence of one or more risk factors that puts them in a higher risk of progression category. We are able to demonstrate in two back-to-back publications in 2024 that our test not only risk stratifies patients that are already at high risk of progression, but also identifies patients whose squamous cell carcinomas are likely to be responsive to radiation therapy or unlikely to be responsive to radiation therapy. Going back to the risk stratification area, we also have our initial lead test for use in invasive cutaneous melanoma called DecisionDx-Melanoma.
If you jump down to the bottom there, also have a test that we acquired a couple of years ago called TissueCypher, which is for use in patients diagnosed with Barrett's esophagus disease and a grade, which is either going to be non-dysplastic, indefinite, or low-grade dysplasia. We can get into the use of that test here shortly. We announced a few weeks ago that we also acquired a second company in the gastroenterology space, this one called Provise, which has a complementary test in the same area as TissueCypher. It also has some upstream pipeline opportunities, which might move us into a diagnostic support area as well going forward. One can organize how one runs a business in different fashions. This is how we choose to organize Castle Biosciences.
I think we first look at identifying a clinical question we can solve through advanced molecular diagnostic tests. We then look to build a robust base of evidence, which will help us penetrate the marketplace both in terms of physician usage and hopefully reimbursement coverage, and then continue to focus and focus and focus on building and expanding and building and expanding in those areas. Just an example of some of the value that we give on a top-line basis. Our DecisionDx-Melanoma test was developed to really answer two clinical questions, which I'll go into on the next slide here. We recently published data from our prospective multicenter DECIDE publication, which did a couple of things.
One of them is that it also demonstrated that physicians use our test clinically to rule out a sentinel lymph node biopsy procedure, which is a prognostic procedure done in melanoma as it is done in other solid tumors. We also demonstrated in this DECIDE study that when our test is used to avoid a sentinel lymph node biopsy procedure, a question for us remains is, are we doing patient harm? That would be that if you're ruling out a surgical procedure that used to be standard of care across the board with the use of a molecular test, we want to make sure that if our test result is used to eliminate that procedure, that patients don't experience an untoward recurrence downstream that's higher than we would potentially predict.
We demonstrated here earlier this year that at least with two-year follow-up data, the DECIDE study demonstrated quite nicely that patients who had a low risk or low likelihood of having a positive sentinel lymph node, therefore avoided that procedure, had 100% recurrence-free survival. No events of recurrences so far. That is going to change, obviously, but it does show you that it is a very, very low-risk population, which is able to safely avoid an interventional surgical procedure. I talked about the radiation therapy data out of the SCC test earlier. Again, we published two back-to-back multicenter U.S.-based studies demonstrating that our test can predict patients who will respond or not respond to adjuvant radiation therapy. Coupled with that was a very nice review of what that does to cost extraction.
At this point in time, there are a couple of predominant radiation therapy approaches used in the Medicare population. This study demonstrated that if you just look at costs of radiation therapy, so the full course of radiation therapy, and our test was used to help rule out or rule in radiation therapy in these patients that are Medicare eligible, the net savings would be $972 million per year to Medicare, which is a great extraction for a single test in a small population. Finally, TissueCypher in terms of the impact of value on patient care. In the case of Barrett's esophagus disease, Barrett's esophagus disease is the only known precursor for esophageal adenocarcinoma or EAC, which is a very, very difficult diagnosis, about a 20% survival rate at five years if you're diagnosed with EAC as opposed to Barrett's disease.
In the U.S., for a number of years, we have stratified patients according to intervention. If they happen to have non-dysplastic disease or indefinite disease, those patients are typically watched, which means coming back in 3-5 years for a repeat upper endoscopy to see if that Barrett's lesion has progressed or worsened. If you're diagnosed with what's called high-grade dysplasia, which is just before the presence of cancer, those patients are universally ablated, usually with radiofrequency ablation with one of the Medtronic tools, but also sometimes with surgical procedures. You have this low-grade dysplasia group, which is just below high-grade by definition. Those patients are probably a bit of a toss-up. Many of them do go on to have their Barrett's disease ablated because, again, if it's not there, it can't progress to cancer.
What we've demonstrated with TissueCypher is that if you go ahead and look at the wealth of data, which is over seven peer-reviewed publications showing the validity of this test, performance characteristics of the test, that we take patients who have non-dysplastic disease, who have the lowest chances of population of progressing to esophageal cancer, and find people who actually have a risk of progression as if they have low-grade or high-grade dysplasia. It is doing a very, very nice job of taking what is a low-risk population, identifying those bad actors to enable them to do what? Be hopefully cured of their Barrett's disease, not progress to cancer. That is important because, again, looking at standard of care today, as a population, non-dysplastic patients represent around 95% of all patients with Barrett's lesions.
Because they are not treated, they now represent the biggest group of patients that advance on to esophageal adenocarcinoma. A great impact in terms of clinical value and clinical care is directed by one of our tests. Let's spend a couple of seconds here on DecisionDx-Melanoma. A busy, busy word salad slide. I appreciate that. A couple of highlights here. If you look on the middle of the left diagram here, what you can see is that we originally developed our test using only molecular factors, that is, gene expressions of 31 genes, 28 of which were linked to the likelihood of progression or recurrence, three genes which are control genes internal to the patient's tumor itself. We developed that algorithm and locked it down back in 2012, so 13 years ago now.
We believed at that point in time that the approach we had taken scientifically got us to the most accurate molecular or biologically appropriate test to predict progression in patients with primary melanoma. We also thought we should go ahead and make sure that as we generate more and more clinical data from a research perspective, can we actually improve the test? A couple of years ago, we said, "We think we have enough research cases here that we should do what? Open up the kimono again and see, one, was the algorithm that we developed in 2012 the best that we can see still in 2019, 2020?
Can we improve the prognostic performance of our tests to allow our physicians to go ahead and get better value? The second question that we asked at the same time was to say, "Is there any enhancement we can do by looking at any clinical features or pathological features that are routinely captured in the order forms that we get that would get us to a more accurate test?" It turns out the answer on the first question was no. We could not find a better algorithm looking at the raw gene data than one that we located in lockdown 2012, which was nice confirmation, I guess, that we had the right algorithm with the right gene set. We did find that by adding clinical and/or pathological factors, we can get to a more accurate result for physicians' use.
What was interesting was that even though progression to a sentinel lymph node, meaning one or more melanoma cells in the lymph node that's sentinel or close to the original primary tumor, is progression just by definition, if we look beyond that and look at the risk of recurrence beyond the sentinel lymph node, what we find is that actually the clinical and pathological factors are a little bit different for those questions, which we didn't think, which was not intuitive to us initially. What we offer today for our clinicians is two test results on one report, which is to say, if you're interested in using our test to rule out or rule in potentially a sentinel lymph node biopsy procedure, then we will provide you with a result from the intergrated i31-SLNB test.
If you're interested in thinking about after that decision is being made, is this patient at low or high risk of progression, then we provide a second result, which is based upon a different mix of clinical factors you can see up there, which is the i31-ROR of risk or recurrence result. That is used very nicely clinically. Now, what does that mean for patient care? On the right-hand side of this slide, you can see two studies that were published in 2024, which are quite interesting. One of them was an ongoing collaboration that we have with the National Cancer Institute CER database program, which is the U.S.'s tumor registry program. There we were able to take patients who were clinically tested with DecisionDx-Melanoma and match them into patients who were captured in the CER database.
Now, that covers about half the states in the U.S., and there's not always perfect reporting, but at least we can match up patients that we tested clinically with the CER registry patient data. Using CER's methodology, they like to do a 3:1 match. For every one patient that we tested clinically that we can find in the CER database, they'll try and find three patients that are matched clinically and pathologically, basically identical to the ones that we know were tested. We have a tested population that we know because we tested them, and then we have an untested population that both of us know because we can't find those patients in the Castle database, but they're in the CER database.
What was demonstrated in the first publication of this collaboration was that if you just look at physicians who were using our melanoma test clinically compared to patients matched for clinical features, pathological features, and socioeconomic features, so zip codes, access to care, etc., if you were tested clinically, and the expectation then is that you had your care adjusted clinically based upon our test results, you had a 29% lower or improved three-year melanoma-specific survival. That's a fantastic gain in patient care with the use of microdiagnostic test. Similarly to that, we also had published or also saw being published an independent study conducted by Cleveland Clinic, Northwestern, Chicago, and Oregon Health & Science in Portland.
They basically took patients in their practices in those institutions where they took patients where they had performance of a lymph node procedure, patients that were node negative, meaning no melanoma cells found in the sentinel lymph node, so they remained either stage I or stage II clinically, just localized melanoma. They said, "We know within each of our institutions, there were some clinicians who were adopting the Castle test and some clinicians who were not." They also matched patients within those three institutions to patients that were tested that were sentinel lymph node negative to those that were not tested, but also sentinel lymph node negative. Same access to care, same treatment pathway, same access to oncology. Similarly, again, what we found was that, again, patients who had a high-risk class two result had the escalation of care.
They were imaged more frequently to pick up metastatic progression, which means they found it earlier. They found it with a lower tumor burden. That's important because, at least in melanoma, we know that our newer therapies, our PD-1 inhibitor therapies, our BRAF, MEK inhibitor therapies can potentially get to cures or at least a much more robust response if you can pick up metastatic progression in melanoma when it's asymptomatic, meaning smaller tumor burden. Again, second study demonstrating that, again, use of our test to adjust care pathways leads to improved health outcomes. These are two great examples of how we try and build value for both improving penetration of our test clinically and also improving access from a reimbursement standpoint.
Another question that most of our clinicians ask us is to say, "Well, we can see data over all stages, stage I and II and III, stage I being thinner tumors, stage II being a little bit thicker, but still located to the primary tumor site, no evidence of progression, and stage III, meaning you have a positive sentinel lymph node." The largest patient population happens to be stage I patients. Those were sort of thinner tumors, which is good because we're diagnosing earlier.
A consistent question we get was to say, "Well, your overall data is interesting, but I really want to understand in the patients I see day in, day out, which is these thinner stage I tumors, how does your test perform?" Again, we had published in the last year or so a couple of publications showing that you can either look at the combined cohort of stage I patients from previously published data, just over 1,200 patients, or you can take a subcut of 5,000 and some odd patients out of the CER database and show excellent performance on risk stratification of these thin patients. I'll switch over to DecisionDx-SCC test. This was the test that we originally developed primarily focusing on patients with cutaneous squamous cell diagnosis of the skin, so SCC of the skin.
Those patients fall into two rough buckets from a guideline standpoint. You have patients that just have a diagnosis of squamous cell carcinoma, usually small, easily curable, no risk factors present. From our perspective, there's no clinical utility in a test that we would develop to help that population. That ends up being about 80% of a million-plus patients. The other 20%, about 200,000 patients a year, we believed when we're diagnosed with one or more clinical or pathological risk factors, they were at a higher baseline rate of progressing. There, within current treatment guidelines, physicians currently have options. They can either just watch clinically. They can decide to go ahead and do something in terms of routine imaging surveillance to pick up spread earlier.
They can consider performing a sentinel node biopsy procedure on that patient like you do in melanoma, but they also have therapeutics like radiation therapy. We initially thought, "Well, you have existing pathways in these people with high-risk SCC. Some patients are getting interventions like adjuvant radiation therapy.
Some patients are getting nothing because the accuracy of the prognosis is not very great looking only at clinical and pathological factors. We were successful in developing a test this time using 40 genes from that patient's primary SCC test to be able to say, "Hey, in that high-risk population, we can find people who are actually quite low risk, similar to the baseline population, but we also find people who are quite aggressive." If you're going to go ahead and decide on, "Is clinical follow-up okay for patient A, but patient B should get radiation therapy today," we can help that clinician, usually a dermatologist, make that call accurately. Fantastic clinical use from our perspective. We aren't creating new pathways. We're just allowing a more accurate assessment of a current pathway.
Along the way, we also looked at the second question, which I mentioned earlier, which was to say, "If adjuvant radiation therapy is the best intervention one has in their toolbox today for SCC patients, can your test help find people who may not benefit or who may benefit from radiation therapy?" We are successful in achieving that goal as well. Today we have here is a risk stratification test, which is to say our overall risk of progression of metastasis runs around 13% in our clinical studies to date. We can take patients from a 13% population average and say, "Actually, we can reduce that by half to about 6.5% likelihood of progression," or at our highest risk class to be result, you can see here it's a multiplier up on the risk perspective.
Those patients, if you look at pathological factors only, look the same, but they obviously have a tumor which acts quite differently, which to me says you would change the care pathway accordingly. Similarly, on the radiation side of the question, what we've demonstrated in our two large multicenter studies published last year is that if you happen to have a Class 2B test result, we can show roughly a 50% reduction in the rate of metastasis at five years for patients who received adjuvant radiation therapy compared to those that did not receive it, or we did not see statistical differentiation between patients with a low-risk Class 1 test result or a Class 2A test result. To put that into a slide viewpoint here, whoops, this is a build slide here. We'll show the, whoops, I guess we won't show the final one there.
In the build slide perspective here, what we saw here was that patients, again, with Class 1 or Class 2A test result, no signal in terms of radiation responsiveness. One could argue an anti-signal, but I think that the noise numbers are too small. If you go ahead and look at a Class 2B test result, you see a fantastic reduction in the rate of metastasis based upon the class call when you take people who received radiation therapy versus those that did not receive it. Looking at time here. Let's skip over to financials here for a minute here. We believe we built a very strong, resilient business over the last four or five years since we went public in 2019.
I'll go through a couple of the highlights here, but we have a few goals this year, which is to go ahead and, one, continue to go ahead and drive robust test volume growth on a pro forma basis, to maintain what we see as industry-leading adjusted gross margin, to achieve operating cash flow positivity by the end of 2025, maintain a strong balance sheet, which is at $275 million at the end of the first quarter, and follow what we believe we have done, which is to say, "Let's take our capital and put it to uses that move the business forward in a very positive way," be it improving or increasing our commercialization investments, improving our pipeline or R&D investments, or finding things that actually fit our strategic focus. I'll go through a few things from the first quarter here.
What you see here on the left and right is net revenue by quarter and test volume by quarter. A couple of things here. One of them is that you can see from the first quarter of 2024 to second quarter, a sort of a reasonable bump up in terms of revenue. We see that typically year in, year out at Castle with seasonality primarily related to melanoma. We expect a similar kind of change, I guess, to be in second quarter 2025. Part of that is seasonality because the second quarter represents the single largest number of working days for our physician base compared to other quarters. We think from a skin cancer perspective, patients are taking off winter coats and hats and gloves and putting on short-sleeve shirts and shorts, and you're just detecting more melanoma or more skin cancers popping up.
There is kind of a combination there, but I anticipate that. We see sort of a roughly flatter or slower growth modality quarter-over-quarter for the ensuing quarter. That is about what we expect to see on a routine basis here, and I think that is what is in our forecast and our guidance for 2025 as well. On test volume, it looks a little peculiar there. You might recall that we made the decision in the first quarter of this year to discontinue our iDgenetix test, which is shown in the purple on the test volume standpoint. The iDgenetix we believe, is the industry-leading pharmacogenomic test by combining both drug gene information, drug-drug information, and lifestyle factors into a single report for patient care.
However, as we looked in the last couple of years at the significant changes in reimbursement challenges in this marketplace, we made the decision last fall to go ahead and move from an in-person, field-based sales and marketing effort to only telephonic-based to watch and see what we could do. They made the decision earlier this year to go ahead and discontinue that test. The volume looks a little funny. Our overall volume of our maintained product, so it has been growing quite nicely. That translates into a couple of nice things. One of them is that if we look at adjusted gross margin, we see that we're maintaining roughly 80%, 81% adjusted gross margin over the course of the last five quarters.
Our operating expenses, you can see the $28.3 million in the first quarter of 2025, that reflects essentially our write-down of the amortization costs of the iDgenetix test from the discontinuation perspective. An anomaly there from that standpoint, the rest of it is fairly consistent as we would expect. If we look at operating cash flow, we see trends that we expect. We do pay cash bonuses in the first quarter of each year. We also front-load some other healthcare expenses for our employee base in the first quarter. As you saw in 1 Q 2024, we used about $6.8 million in cash to operate the business. You saw in first quarter of this year, again, about a $6 million use of cash to go ahead and fund those first quarter periodic expenses.
Then we picked up last year roughly $23 million-$24 million in operating cash flow each of the three quarters, which to us reflects as growing into a very nice business over the last four or five years, especially last year. If you look at adjusted EBITDA by quarter, it is the exact same kind of trend there. As a final slide here to come through, we believe that we have taken a thoughtful approach as we use to invest our capital in the marketplace. We went public in 2019, I think did two raises in 2020. Is that right, Frank? Since that time, have not had to go and go back to the capital markets. We have increased our balance to $275 million in cash, cash givens and investments through the first quarter of 2025.
We use that to fund three areas of the business, I would call it. One of them is make sure that we're optimizing our commercialization investments. From a dermatology perspective, we are sitting in the kind of low 80s from a sales representative perspective in territory salespeople. We think that's probably about where we need to get to. There are a few territories that are approaching $3 million in sales, and we will look to split those territories up into two. That's too many ongoing customers to let our sales reps maintain a balance between growing within the customer base and growing new customers. That's about our cut point, is about $3 million is sort of the concerning factor on current volume.
In the GI side of the business, which has a higher growth potential right now, we went from 24 sales representatives a year and a half ago to sort of the mid-60s the first part of this year. That will help to go and drive penetration in that Barrett's esophagus gastroenterology marketplace quite nicely. That is the other area of commercial optimization. From an R&D perspective, we're split between sort of supporting our currently marketed products in dermatology, gastroenterology, ophthalmology, and also pipeline investments going forward. We are laser-focused on a lesser priority in general, which is to say, are there assets that we can pick up that will actually make our value to dermatologists, our value to gastroenterologists stronger? That may be things that we can market today. It may be pipeline tests we can market tomorrow.
We have certainly been open in the past looking at saying, is there another area or so that would actually strengthen the foundation of the company from a performance standpoint that we should consider going into? Again, a very disciplined approach. Thankfully, we're in a position today to make calls that are good for Castle in the mid to long term as opposed to being too reactionary in the short term. With that, I think we're done with the core presentation here and have three minutes left for any questions. Frank, do you have a question for me? No questions from the CFO either. All right. Thank you very much.
Thank you.