Earnings Call: Q3 2019

Oct 22, 2019

Good afternoon, ladies and gentlemen, and welcome to Veracyte's 3rd Quarter 2019 Financial Results Conference Call. As a reminder, today's conference call is being recorded. I'd now like to turn the conference over to Keith Kennedy, Veracyte's Chief Operating Officer and Chief Financial Officer. You may begin. Thank you, Sydney. Good afternoon, everyone, and thanks for joining us today for a discussion of our Q3 2019 financial results. With me today are Bonnie Anderson, Veracyte's Chairman and Chief Executive Officer Julia Kennedy, our Chief Scientific Officer and Chief Medical Officer and John Hannah, our Chief Commercial Officer. Before we begin, I'd like to remind you that various statements that we make during this call will include forward looking statements as defined under applicable securities laws. Forward looking statements include those regarding our future plans, prospects and strategy, financial goals and guidance, product attributes and pipeline, drivers of growth, expectations regarding reimbursement and other statements that are not historical fact. Management's assumptions, expectations and opinions reflected in these forward looking statements are subject to risks and uncertainties that may cause actual results and or performance to differ materially from any future results, performance or achievements discussed in or implied by such forward looking statements, The company can give no assurance they will prove to be correct and will not provide any further guidance or updates on our performance during the quarter unless we do so in a public forum. Please refer to the company's October 22, 2019 press release and the risk factors included in the company's filings with the Securities and Exchange Commission for a discussion of important factors that may cause actual events or results to differ materially from those contained in our forward looking statements. Prior to this call, we announced our Q3 2019 results, which are available on our website atveracyte.com under Press Releases in the Investor Relations section. We also published a financial presentation, which I will reference during my remarks and a presentation on the preliminary data for our nasal swab classifier, which Bonnie and Julia will reference later. These presentations are also available on our website under Events and Presentations in the Investor Relations section of our website. I will now turn the call over to Bonnie Anderson, Veracyte's Chairman and CEO. Thank you, Keith, and thanks, everyone, for joining us today for our Q3 2019 earnings call. We are speaking to you from the CHEST conference in New Orleans, where we announced exciting new data this morning regarding our nasal swab classifier for early lung cancer detection. We'd like to address and underscore 3 key points on today's call. First is the continued strength and momentum of the business. 2nd is the significant progress we are making in our pulmonology franchise and third is the new data for the first ever nasal swab test, which we believe will be pivotal to us scaling our business and expanding to global markets. We'll start by reviewing our Q3 results. We generated record revenue of $31,000,000 in the Q3 of 2019, an increase of 32% and genomic test volume of 9,941 tests, an increase of 24% over the Q3 of last year. At the same time, we improved our net cash used in operating activities to $1,600,000 an improvement of 13% over the prior year. We reiterate our full year revenue guidance of $119,000,000 to $122,000,000 and net cash used in operating activities of $2,000,000 to $4,000,000 and remain on track to reach operating cash flow breakeven before the end of 2019. I will now share updates on the key metrics with which we are measuring our performance in 2019. The first is revenue growth. We continued to see strong growth across all three of our core classifiers. In pulmonology, we delivered over $1,000,000 in Percepta revenue for the Q2 in a row. Customer response to the Percepta Genomic Sequencing Classifier, or GSC, which can now down classify patients to low risk as well as up classify them to high risk following an inconclusive bronchoscopy continues to be quite positive. We are on track to report an estimated 3,000 results this year, doubling the prior year results. Our Envisia classifier is gaining impressive traction as well with its ability to help physicians distinguish idiopathic pulmonary fibrosis or IPF from other interstitial lung diseases or ILDs without the need for surgery. Over 135 sites have to report Envisia test volume between 5,101,000 tests for the year. We also continued the strong momentum of our flagship Afirma business in thyroid cancer, where we reported 3rd quarter test volume of 8,925. This brings our year to date Afirma volume to over 26,000 tests, which is a 21% increase over the same period last year. Our solid product growth reflects the continued success of our integrated sales strategy. In fact, the number of accounts that are using more than one of our products grew by 30% in the 3rd quarter to a total of nearly 220 accounts. Finally, in addition to product revenue, we booked $4,300,000 in biopharmaceutical revenue during the Q3. This stems from our accelerated milestone achievements with the Johnson and Johnson Lung Cancer Initiative as well as our continued collaboration with Loxo Oncology. Evidence development is our 2nd metric of success. Here too, we experienced significant progress in the Q3. In September, 2 new papers added to the growing body of evidence supporting the use of Afirma Expression Atlas or XA to help guide surgery decisions in patients whose thyroid nodules are deemed suspicious for cancer by our Afirma genomic sequencing classifier. This includes a strong clinical and analytical validation paper published in Frontiers in Endocrinology, as well as a study published in the journal Thyroid reinforcing the test value in identifying the presence as well as the clinical relevance of specific gene alterations found in preoperative thyroid nodules. We are also looking forward to the American Thyroid Association's annual meeting next week where multiple poster presentations will showcase data derived from the Afirma XA further defining the genomic In addition to the new nasal swab data, 5 studies for our commercial products are being presented. These include 3 abstracts showing that the Envisia classifier enhances physicians' ability to confidently diagnose IPF in combination with high resolution CT imaging as well as 2 abstracts demonstrating the clinical validity and utility of the Percepta classifier in lung cancer diagnosis when bronchoscopy results are inconclusive. I am especially proud of the strong scientific and medical reputation our team is building in pulmonology. Our 3rd metric of success is pipeline advancement. In addition to our nasal swab classifier, we are announcing that we formed a research collaboration with National Jewish Health, one of the leading respiratory centers in the country to explore opportunities to further improve diagnosis of IPF and other ILDs. Through the collaboration, we will combine diagnostic imaging data from National Jewish Health with whole transcriptome RNA sequencing genomic data from our rich biorepository of ILD patient samples. Our goal is to determine if these expansive and complementary data sets, when informed by our deep machine learning expertise, can enhance diagnosis across the care continuum for ILD patients. We look forward to keeping you appraised of outcomes from this collaboration and anticipate it being showcased at a future scientific meeting. Our 4th measure of success is financial discipline. Here too, our team has continued to excel. Our net cash used in operating activities for the Q3 of 2019 was $1,600,000 a 13% improvement over the Q3 last year. We remain confident that we will achieve our goal of reaching cash flow breakeven before the end of 2019. I will now turn the call over to Keith to review our financial results for the Q3 of 2019. Thank you, Bonnie. As mentioned earlier, our Q3 2019 financial presentation is available under Events and Presentations in the Investor Relations section of our website. Turning to Page 3 of our financial presentation. Our performance against 6 financial key performance indicators or KPIs for the Q3 of 2019 as compared with the prior year's quarter, including select highlights for each metric at the bottom of the page are as follows: Revenue of $31,000,000 increased 32%. Excluding $4,300,000 of biopharmaceutical services, revenue of $26,700,000 increased 15%. Genomic volume of 9,900 and 41 reported tests increased 24%. Gross margins of 71% increased 600 basis points. Excluding biopharma services, gross margins increased 200 basis points from 64% to 66%. Operating expenses, excluding cost of revenue, increased 21%. Net loss of $700,000 improved 84%. Net cash used in operating activities of $1,600,000 improved 13%. And at September 30, we had cash and cash equivalents of $196,000,000 Turning to Page 4 of the presentation. Our performance against these 6 KPIs for the year to date period ended June 30, 2019, compared to the same prior year period shows strong comparable performance. The next six pages outline the sequential and year over year results underlying each of the 6 financial KPIs. A few observations. As illustrated by the revenue and genomic volume trends on Slides 56, we continue to see positive momentum across the business. Our lung portfolio represented approximately 1,000 tests or 10% of our genomic volume this quarter. Turning to Page 12 and our 2019 guidance. As Bonnie stated earlier in her remarks, we are reaffirming our revenue guidance of $119,000,000 to 122,000,000 dollars and net cash used in operating activities of $2,000,000 to $4,000,000 In the 3rd quarter, our loss from operations was $1,800,000 which included $3,600,000 of depreciation, amortization and stock based compensation. To add some additional color on our outlook for 2019, principally in the Q4 of 2019, In Q4, we anticipate receiving $4,000,000 in payments from Johnson and Johnson related to milestones achieved in the second and third quarter. We expect gross margins, excluding the impact of biopharma services revenue, to be within the 65 percent to 67% range. In Q4, we expect our average quarterly spend for sales and marketing to stay within $1,000,000 band around the average quarterly spend of $13,500,000 and our average quarterly spend for our combined G and A and R and D spend to stay within $1,000,000 band around the combined average quarterly spend of $10,500,000 I will now turn the call back over to Bonnie. Thanks, Keith. Now we'll turn to our big news of the day. The preliminary data that we announced earlier today for the first ever noninvasive nasal swab classifier to enable early lung cancer detection and diagnosis. I'll first walk you through the market and where in today's clinical pathway of care we are positioning this test. Doctor. Julia Kennedy, our Chief Scientific and Medical Officer, will then discuss the test development, these exciting new data and our next steps for bringing the test to market. As Keith mentioned, we have published a presentation, which we'll reference in our remarks. You can find this in the Investors section of the website under Events and Presentation. The first five slides of the presentation give you background on how Veracyte is transforming care throughout the patient journey, Starting with our mission of improving diagnostic accuracy, we have expanded to advance early detection and informed treatment decisions. We see our nasal swab classifier as a key opportunity to move upstream in the patient care continuum. Turning to Slide 6. As you likely know, lung is the biggest cancer killer in the U. S. And worldwide, and early detection of lung cancer is key to saving lives. A patient's odds of survival increase significantly when the disease is caught before it has spread. As illustrated on Slide 7, lung nodules are often the first sign of lung cancer. But since most nodules are benign, a big challenge for physicians is knowing which patients need invasive biopsies and which patients can be safely followed non invasively. Currently, about 2,000,000 lung nodules are detected by imaging in the United States each year. These cases are derived from the approximately 10,000,000 patients who are at high risk for lung cancer and are thus eligible for annual CT screening as well as approximately 1,600,000 cases where lung nodules are detected incidentally via CT scan or x-ray. We expect the number of nodules to increase as screening programs continue to expand. When lung nodules appear suspicious on CT scan, patients are typically referred to a pulmonologist for workup and diagnosis. The pulmonologist then decides how to work up each patient based on the risk that that nodule is cancerous. Those assessed at being higher risk typically are advanced to more invasive diagnostic evaluation and potential treatment, whereas those considered low risk are monitored non invasively. Today, there is a lack of standardization and objectivity in determining this risk. As a result, many patients, many benign patients undergo workups they do not need and patients with cancer potentially experience delayed diagnosis and treatment. We propose our nasal swab classifier will help solve this problem. As you can see on Slide 8, we are positioning our nasal swab classifier to provide pulmonologists with a simple, convenient and objective tool to inform patients' diagnostic and treatment pathway. The sample for this test can be collected right in the pulmonologist's office with no need to refer the patient out for a blood draw. And as Julia will show you, the preliminary data suggest that our nasal swab classifier can transform this trajectory for many patients at this point, enabling patients with lung cancer to get the treatment they need sooner, while helping patients whose nodules are actually benign avoid unnecessary and costly invasive procedures. By accurately identifying low and high risk patients, we believe we can greatly reduce the number of patients in the current intermediate risk or uncertain path forward risk pull. This would significantly improve patient management and health outcomes compared to today's current standard of care. On Slide 9, we note that we estimate that the market opportunity for our test is approximately $1,900,000,000 in the U. S. This is based on the estimated 750 1,000 patients annually who are referred to a general pulmonologist for a workup of a suspicious looking lung nodule found either through lung cancer screening or incidentally. We estimate the market opportunity to be approximately 3,900,000,000 dollars for both the U. S. And the European Union. As a reminder, on Slide 10, development of our nasal swab classifier stems from our long term collaboration Johnson and Johnson Innovation and the Lung Cancer Initiative, which we announced in January 2019. We also plan to explore opportunities to deploy our field of injury technology at other points along the lung cancer care continuum. We believe the overall global lung cancer diagnostic market is approximately $30,000,000,000 I will now turn the call over to Doctor. Julia Kennedy, our Chief Scientific and Medical Officer to walk you through the new data. Thanks, Bonnie. As Bonnie mentioned, our nasal classifier is built on proven field of injury science. This means we can detect genomic damage associated with lung cancer in the airways of current or former smokers. Our Percepta GSP uses this approach to determine from cells collected in the main bronchial airway whether a remote lung nodule is likely cancerous or not. In a study published in the Journal of the National Cancer Institute in 2017, researchers from Boston University showed that this genomic damage could also be detected in epithelial cells collected from the nose. Our preliminary data confirms and extends these findings allowing us to develop a prototype set of biomarkers for a new nasal swab classifier. Our methods are shown in Slide 11. In order to develop and assess the performance of our preliminary classifier, we're very fortunate to have access to nasal swab samples collected from our large scale multicenter perspective AEGIS I and AEGIS II clinical studies. We acquired the samples through our 2013 acquisition of Allegro Diagnostics. These studies recruited thousands of patients who were undergoing evaluation for potential lung cancer and who had long term follow-up. In addition to final adjudicated benign or malignant diagnoses, we have extensive clinical and radiology information for each sample, such as age, gender, smoking status, use of inhalant treatments and nodule characteristics found on imaging. Importantly, having this rich repository has enabled us to shave several years off of the time it would otherwise take to develop a test like ours. As shown on Slide 12, we begin by randomly dividing the Aegis cohort into a training set of 411 nasal samples and an independent test set of 261 samples for our early biomarker discovery. You can see the broad range of clinical features and nodule characteristics that were included in both sets. We extracted RNA from the nasal samples and conducted whole transcriptome sequencing to measure a quarter 1000000 features. We then developed machine learning models that were trained to identify 2 conditions, benign and malignant cases using the genomic and clinical features. We developed and evaluated hundreds of models using nested cross validation. Several of the best models were selected and then used to score an independent test set, meaning these samples were not involved in training the algorithm. Data for the best performing model, which provides us with a firm foundation for developing our future nasal swab classifier, were presented earlier today at the CHEST meeting. Slide 13 gives an overview of our prototype classifier. The model, which is a penalized logistic regression classifier, was developed by machine learning on clinical features such as those relevant to smoking history as well as other clinical features and combine them with genomic features extracted through our RNA sequencing pipeline. The foundation of this capability has been built over the last decade at Veracyte. Turning to Slide 14. When we initially assessed the test, we took a common approach of selecting a single cutoff to distinguish benign from malignant cases. We found that test sensitivity for lung cancer was 97% and its specificity was 46%. This means that among patients whose nodules were actually benign, the genomic test classified over 40% as low risk for cancer with high accuracies so that these patients could be monitored non invasively. High sensitivity is crucial for a test used at this point because if physicians are directing test development and was what we included in our initial abstract submission to CHEST. In analyzing the data further, we saw that in addition to seeing high performance in classifying samples as low risk, high risk cutoff to improve the test specificity or malignancy. As you can see on Slide 15, we subsequently demonstrated the test performance using 2 cutoffs. In addition to the previously shown high sensitivity, which allows us to call a substantial portion of the true benign patients as low risk, while missing very few cancers, we found that among patients whose nodules were malignant, the test classified over 40% as high risk for cancer and had a specificity of over 94%, meaning these patients could be directed to more invasive diagnostic procedures and treatment with a low rate of false positive results. Thus, the test achieved success at both ends of the spectrum, calling truly benign patients low risk with high sensitivity and calling truly malignant patients high risk with high specificity. We also observed that the test performance was consistent regardless of lung nodule size or location as well as cancer subtype or stage, giving us confidence that the test can perform in smaller nodules and a diversity of cancers that might be encountered by physicians. Now let's look at Slide 16. Because the cancer prevalence rate in our AEGIS cohort was high, 78%, we modeled how the test would perform in a population with a cancer prevalence of 25%, which is more aligned with the patients on whom it will be used. In this scenario, with its sensitivity of over 95%, the test negative predictive value would be 98% when it identifies patients as low risk with fewer than 5% of truly malignant patients misclassified as low risk. And its positive predictive value would be 76% when it identified patients as high risk for cancer with less than 6% of true benign patients misclassified as high risk. We're very pleased with the strength of these early data and look forward to further refining the class the of the test system and sample collection methods, along with additional studies needed for our publication pipeline. I'll turn the call back to Bonnie. Thanks, Julia. To conclude, we are very excited about the progress with our nasal swab classifier and truly think it can be a game changer in the fight against lung cancer. As we have illustrated on Slide 17, we believe our noninvasive nasal swab classifier can guide patient workup consistent with guidelines to avoid invasive procedures on benign nodules, while making a more timely diagnosis and treatment decision on high risk patients. And we believe this will be a tremendous tool for physicians in standardizing care with objective measures. Additionally, this test will greatly reduce the pool of intermediate risk patients who will follow the course of standard treatments as they do today. We expect to finalize the test over the coming year and commercialize it in early 2021 in the U. S. In closing, we are thrilled with where Veracyte is today and where we are heading in all three of our clinical indications. We expect to provide our 2020 guidance when we report our 4th quarter 2019 financial results. But as we look ahead to next year, we expect to deliver revenue, excluding biopharma services and genomic test volume growth of over 20% in 2020. It should be another exciting year. I will now ask Sydney to open up the call for questions. Thank you. Our first question comes from the line of Sung Ji Nam with BTIG. Your line is open. Hi, thanks for taking the questions. So maybe starting out with a clarification, I think you mentioned earlier, early detection versus diagnosis, but this nasal swab classifier, when available, you will wouldn't require additional workup post the results. Is that correct? So you can potentially recommend okay. Yes, that's correct. Although with this test, you can avoid the multiple steps that sometimes takes place today and we'll be able to get the high risk for cancer patients in for a diagnosis immediately on treatment and avoid a workup on patients that are truly low risk because our results are near, you could almost call them benign with the level of performance we have. So by cutting those 2 pools in half, we are going to be able to move half the patients to the next step they need, and that's a pretty exciting result. Okay. That makes sense. And I haven't had a chance to look at the publication cited here, but could you kind of go over more in more detail the intended to use population? Is that based on specific medical guidelines or medical society guidelines that we should be aware of? The way this works today is we know there are about 10,000,000 patients at risk that are currently eligible actually to get a low dose CT screening for lung cancer free of charge. That was part of the ACO requirements that came out a couple of years ago. Unfortunately, though, many of those patients that are at risk and are eligible for screening do not get in for screening. And what we hear and what our data show is that often that's because of the concern about the workup required today to make the diagnosis being highly invasive, costly and risky. So the first thing is we believe we are now going to help improve this pathway of workup so that these 10,000,000 patients that are eligible get in for screening. We think that the imaging low dose CT screening is actually a really good tool. It's sensitive. The challenge with it is the false positives. And that's what nasal risk can greatly improve on. In addition though to those screening patients, there are about 1,600,000 patients today that present with nodules just through incidental findings. These patients may be undergoing a workup for before they go into surgery. They might have a CT scan after following a car accident or an x-ray. And when these incidental nodules are found, then they are determined workup or not workup. Out of all of those nodules found, about 2,000,000 today in the U. S, there are a number of those nodules that are determined very, very low risk, so they're not even referred for workup. But about 750,000 of those nodules will be suspicious enough to be referred for workup. And so when you think about what could happen next with those 750,000 patients, And this is about a 25% prevalent group. So there's roughly 200,000 cancers detected every year in the U. S. Some of these numbers round a little bit. So if our nasal test were to be performed next, we would be able to take over 40% of those 550 benign patients and at that point in time classify them as low enough risk to not need workup. At the same time, we would be able to take about 40% of the 200,000 malignants or about 80,000 cancer cases and put those patients into a high risk for cancer bucket so that they move through the process quickly, get diagnosed right away and on treatment. That's the magnitude of what we're going to be able to do here. In the meantime, we're going to end up with still an intermediate risk group bucket because we do fantastic on the bookends, but you'll have the group in the middle that are still intermediate risk and that bucket will be about half what it's estimated to be today. So a tremendous improvement and an objective improvement, where when you look at the guidelines and they recommend patients with a low risk be moved to CT follow-up, our nasal classifier low risk call achieves that less than 5 percent risk of malignancy. So that's excellent. That fits right in with the guideline today. And obviously, with a positive predictive value estimated at over 75%, those patients want to get diagnosed and on treatment for lung cancer. So it ought to have a great impact for patient care and hopefully get more and more patients into screening for lung cancer. That's super helpful. And then lastly from me, if I understand this correctly, I think this is the same patient sample cohort that have bronchoscopy swab data as well. So I was wondering, it might be too early, but how I guess, how does how do the results that you presented compared to Percepta performance? I'm not sure if this is the right way to think about it. Yes. It's a different intended use population, right? Because with Percepta, we are capturing the patients that have gone through a bronchoscopy workup, about 380,000 patients today undergo bronchoscopy workup and up to 50%, 60% of those are inconclusive. And that is the group we're measuring with Percepta. So it's a completely different subset of the patient population and the prevalence is actually a little higher. I think when we presented our Percepta data, the prevalence of cancer in the testing population there is about 30% to 35%, I believe, somewhere in that neighborhood. So it's a little bit higher than what this all comer group will be. And also many of the suspected cancer patients that they're defined by more than 60% risk of cancer as assessed clinically would often go on to the surgical biopsy in place of bronchoscopy workup. So we are fortunate that when the AAGES clinical trial was conducted that they had the foresight to gather nasal swabs on all these patients, so that we could go back to this library of very well curated samples and begin the work in the nose. Of course, we obtained all of that when we acquired Alevra. I think in the script, we said 2013. It was actually 2014 that we made that acquisition. And Percepta was very far along and already validated, so that it was smart to take that to market first. And now we're able to turn our attention to what we can do in the nose. And we believe there are even more opportunities beyond the one that we are presenting the data today, where this test might add value, in the global marketplace for improving lung cancer diagnosis and treatment. Great. That makes a lot of sense. Thank you so much. Thank you for joining us. Thank you. Our next question comes from the line of Thomas Flaten with Lake Street Capital Markets. Your line is now open. Thank you. Congrats guys on the nasal swab data. It looks great. Just I'm looking at Slide 17 and I was hoping Bonnie if you could help us understand the future of Percepta in this new paradigm. I'm assuming it obviously fits into that intermediate group, but if you could add some color to that, that'd be great. Yes. I think we would expect that the intermediate risk group would follow the workup as they would today, which is typically bronchoscopy is very attractive at this point because it's less invasive and this population has a little bit lower like a moderate risk of malignancy. In the numbers that we were talking about earlier, it would be about 25% to 30% risk of malignancy after the nasal test is performed. But what you also have to keep in mind is with the nasal test intercepting the classification of the patients in advance of workup, we expect the funnel of patients coming through and nodules being found to grow. So while that intermediate bucket will still be the likely bucket to go on to bronch, we also believe that the 700 50,000 patients that are moved through the workup could double or triple as the 10,000,000 patients at risk for lung cancer get screened. So the dynamic will change. The market for both tests should actually grow over what it is today. And we believe with the combination of our nasal swab classifier, the Percepta GSC and eventually Percepta Xpression Atlas that we are going to be able to inform on everything from early detection to diagnosis to treatment decisions at the time of diagnosis with the partnership with the pulmonologists that are making these decisions. And we're really excited about the success and the advance that we have made in our pulmonology franchise. That's great. Thank you. Keith, one for you with respect to guidance. There's an increase above and beyond what you had projected coming out of the Q2 in the biopharma revenue from I think you had $10,000,000 pegged in the second quarter and now we might be up as much as $6,000,000 on top of that. Can we infer then from that that there's an implied drop in the guidance that's associated with testing revenue? We've been updating you quarterly. I think this quarter there's $2,000,000 of incremental service revenue over where we were last quarter and that was due to advances that we've made in Percepta milestones in the nasal outcomes as we come out with this test. So we are relative to the street, I think we're probably $1,000,000 total below in revenue on the classifiers and $1,000,000 up on service revenue in total. So I mean, it's there's a little bit of a difference, but it's not a market difference. Okay. Thanks for taking my questions. Thank you. Thank you. And our next question comes from the line of Puneet Souda with SVB Leerink. Your line is open. Yes. Hi, Bonnie, Keith and Julia. So question that I have is if I could ask on the data, if you could help me understand your rationale for 2 cut classifier here in the first place. I know you elaborated a bit on that, but which one on the either end has more value in your view? Is it the lower risk or the higher risk? And I'm asking that because the specificity is lower and the low risk and sensitivity is lower in the high risk group. Maybe just if you could take a minute and help us educate on those front and sort of what I'm trying to get to is even after this classifier, what's the group of patients that we are that the assays are still missing and still not part of the ones that are going to bronc. If you could help me understand that aspect? I'll start and then I'll hand it over to Julia. But I have to say, when we can take nearly for over 40% of true malignant patients and classify them as high risk with very high specificity, which means there's very, very few benign patients going to be put into that class, is not much better improvement to that. On the low end, it's the same thing. Being able to take 40% of the patients that would be in a pool of intermediate risk patients today that the physicians would sort of pick and choose and calculate and try to decide the best next steps, when we can put half those benign patients or over 40% of those patients into a low risk bucket with over 96% sensitivity, we are doing that and not missing hardly any cancers. So I think the first thing that's important to clarify is we're not missing anything here. We are taking this pool of all these patients with nodules and putting nearly half the high risk into one class, that's obvious what you do next, nearly half into low risk, which is obvious what you'll do next and cutting that intermediate risk pool literally in half. Those patients will follow whatever you would do today with the entire bucket. Julia, do you want to talk about the rationalization and rationale for this 2 class approach? We realize this is kind of cutting edge, sort of new, but it's pretty exciting. Yes. I would say it's a bit of an unconventional approach, but what we've done is play to the strengths of our data at either ends of the spectrum. So it's very hard to get a test that's both highly sensitive and highly specific. And by adding these two cutoffs, we basically play to the strength of the high sensitivity cutoff at the bottom of the range to call patients low risk with very high sensitivity. And then at the top of the range, we're able with the second cutoff call those with very high specificity to be truly malignant with very, very low rates of false negatives on the bottom and false positives up at the top. It's just really playing this to the strength of the test. Okay. That's very helpful. If I could ask on the pricing of the test, Bonnie, the range here is $800,000 to $2,500,000 Could you elaborate on what was the pathway to get to that pricing? What are some of the benchmarks used for that? And a follow-up there is in past you have created the strong clinical evidence and published that in high impact papers like New England Journal of Medicine and others for Afirma and Percepta. So do you think you have an adequate data here I mean adequate data at this point to go to publication and do such an high impact publication? Or do you need more further studies here to bring that clinical evidence to market? Yes. So it wouldn't be the first time. I bet when we developed Afirma, we actually had the first publication that came out that got referenced many times after the test was launched was in some of the early work that was done. It was 2 to vol and all. And it was a very high impact publication even though it was an earlier data set than the final released product. So yes, our scientists and obviously all of our clinical investigators that are part of this journey with us, I'm sure will want to get this data out in publication. I'd be very surprised if this would not warrant a high profile journal. In terms of pricing of the test and more details on that, we put the range in the market slide just to be transparent that we believe pricing will need to be different perhaps in the EU than the U. S. Part of the market. And we will need to continue to do some work on the modeling around the true value the test will deliver, and then we'll pick a price point that makes sense for the value we're delivering and the market that the test will be sold in. But that reference isn't necessarily because we don't have any idea where we'll price the test. It's more to give you the ranges around how we came up with the market data. But you'll hear more of that as we move. That's part of what we'll be doing over this next year is the market development work, the premarket and commercialization planning and tightening down all of those details, and we'll keep you all very abreast of that progress. Okay. And if I could just ask on the test performance, how much of that is related to the genomic evaluation relative to other inputs in the algorithm, the age of the patient, the pack years of smoking, the nodule characteristics, etcetera. So what is how much is that contributing to the assay? Yes, it's a really good question. I'll ask Julia to that was in the poster actually that was presented. And we'll try to get this poster access to the poster with a link at some point after we get home from CHEST on our website as well. But Julia, why don't you talk a little bit about this versus these clinical calculators? Sure. So, we developed the model to include in addition to gene transcript features from 26 genes, we actually also have what are called interaction terms with those genes and clinical factors. As part of the Aegis clinical cohort, we collected a large number of clinical factors and we use these as features in the machine learning. And so our algorithm actually is a combination of genomic features and clinical features. When this classifier is compared to a commonly used clinical risk predictor, which doesn't have any genomic features and it's just some clinical factors. For example, the Gould model was what we have in the poster. We find that there is substantial improvement of our classifier specificity side. So it increases the number of benigns that are called low risk by 70% over what the gold classifier would do by itself and 18% of the malignant to high risk more than what the Gould classifier would do. Thank you, Julien. Anything else to meet? If I could, last one for Keith. I don't want to miss him. In this gross margin, Keith, expectations for any improvements on that? It's sort of flat lined in the last three quarters. Yes. Well, yes, that's right. 64% to 66% is where we're holding on the margin. We obviously have the lung products are reimbursed at a lower rate than Afirma, obviously. So that weighs on the margin. So we're trying to hold it in that 64% to 66% range for the 1st couple of years as we go down the managed care journey on those products. All right, great. Thank you, guys. Thank you very much for joining us and for the questions. Thank you. Our next question comes from Paul Knight with Janney Montgomery. Your line is open. Bonnie, as we start to think about the commercialization of the product, you're seeing it starting in 2021. I'm assuming that you're kind of thinking that that's the initial days perhaps of private pay and dependent upon CMS approval, you would need published papers. So I'm assuming you're thinking what CMS approval sometime in 20 21 as well and when do you think commercial pay could develop? Yes. Paul, we know it will take a little bit of time. History would show that it can take a year depending on how quickly you're able to assemble the required evidence and get it accepted into publication, etcetera. But we've done this a few times. So I think that you can expect over the next year, we'll probably get as much of this lined up as we can and try to be as quick post commercialization to coverage as we can be. And we'll keep you updated as we have any more clarity on that pathway. And then Keith regarding this, you're going to have about $8,000,000 or so of service here in the second half. That seems to be above, I think, where you were at the beginning of the year. So what's within the core, the diagnostic franchise today, is it it seems like it's a little lower than we started out with the year? Is it pricing? Is it ramp up on tests? What would you give us color on regarding that change in the build out of the model? Well, we've done about 26% growth in revenue year to date on our molecular test. So that's just a firm of Percepta and Envisia. So that's up about $15,000,000 for the 1st 9 months of the year. Our cytopathology business, which we've always talked about as being a flat to no margin business, but an important factor in terms of Afirma and the market for the commercial team, That's down about close to $2,000,000 year over year. So that's dragged from 26% molecular testing down to 20%. And then on the quarter, we probably had about 3% reduction in that growth rate because we had in the prior year revenue recognized for tests performed in prior periods. We didn't have the advantage of this period, so that's where it diluted the growth rate. But by and large, Affirma is growing around 17%, 18% in volume and revenue, and we're accruing around $2,800 for accrued samples on the test, and that's been consistent quarter over quarter. And that's very close to where we predicted Afirma to be for the year. We're growing genomic volume 28% over the year, so for the 1st 9 months relative to the prior 9 months and 24% this quarter over the prior year quarter. And so if you think about our lung test is now 10% of our portfolio, there's probably 6% to 7% growth embedded in the reduction in what we get for that test. So we're getting $1300 and we should be getting $26,000 $2,700 on average for those tests. We factor in patient pays and all that. We're leaving 6% 7% of that growth on the table. And over the next 2, 3 years, we really build that volume up and we get to that commercial journey. That's what creates that tailwind you saw on Afirma. So eventually, our revenue growth rate will exceed our volume growth rate. But in the early years of the product, our volume growth rate is higher than our revenue growth rate, which we actually think is a very positive indicator for the long term growth of the business. Does that make sense? Yes, very helpful. Okay, great. Thank you. Thank you, Paul. Thank you. And our next question comes from Brian Weinstein with William Blair. Your line is open. Hey, guys. Sorry for the background noise. Julia, for you, can you just go back and describe in a little more detail the current standard of care for classification of nodule model, how that's used and just make sure that we understand how your product compares to the performance of that model, which I believe is what this is all based on. I didn't see that in those slides, but it was a big part of the poster that we came down to see today. Sure. The clinical models that are used, and specifically the gold model uses factors, simple factors such as age and smoking exposure, never current or former nodule size and things like that. Our and these variables actually are clinical variables are themselves quite variable and there's a lack of generalizability to these models. So when they're developed in certain cohorts, they don't always generalize to other cohorts. And so there's a lack of standardization, lack of generalizability. What we've done is we've taken clinical factors. We've interacted them with the gene expression terms in the classifier and we've made the test more standardized and less objectivity. So regardless of what the pretest risk that, that nodule may have going into this testing population, we can take those brush nasal brushings from those patients regardless of what their risk is and assign them with this nasal swab test a highly accurate low risk and high risk assessment of their risk of cancer. Yes, Brian, to add to that, we have a lot of market data going out and asking physicians what they actually do today. And what we have found is that while these risk calculators are great tools and they can be used multiple different ones by some of the different physicians. What we notice and what the data show is that regardless of the calculated risk, there is a vast lack of standardization in what is done next. So even though they may use the tool and try to assess the pre workup risk, in reality, you have many low risk patients undergoing aggressive biopsy for diagnosis. You have more higher risk patients that actually are moved to watchful follow-up because of the lack of objectivity and standardization across the universe of sites. So we think this is a really great tool to help sort of set a new standard of care of how risk is actually assessed pre workup. Great. Thank you for that clarification. And then last one for me is just moving further upstream with this technology and the opportunity to move to maybe more of a true screening population. Talk about the opportunity there, how you guys are thinking about that and how that might contribute to the $30,000,000,000 overall market opportunity that you guys have referenced? Yes. Well, I think it's pretty provocative to be able to detect genomic damage in a nasal airway swab and develop this level of accuracy in a very large percent of these patients. So it is a pretty provocative and pretty exciting data from that standpoint. If we can do that post nodule detection, there certainly is no reason to imagine that something is magical about that patient with the nodule. So we're definitely interested in ways that we can move further upstream. It would certainly be interesting under the new paradigm and thinking in the pharma world to be able to predict precancer with patients that you could halt progression of disease perhaps. And certainly, those are some of the discussions we have with our J and J collaborations. But I do want to come back to a point that we actually think is pretty important, and that is that screening broad populations for lung cancer is in fact, one of the posters right next to ours today showed that screening lung cancer screening with low dose CT can be very effective, but even today, it's expensive because of the workup costs associated with what you do once you find those nodules. We believe that it's highly unlikely that broad population screening will ever be done in lung cancer, because there are factors that can be used and that have been used to identify the at risk population. And when you have an at risk population, you're really not doing population screening. You're using a tool to get those at risk populations to get detect the cancers that are there early. So some of this is semantics. But we do believe that low dose CT then having an inexpensive low radiation dose screening tool in place that is completely noninvasive is not at all a bad place to start for at risk lung cancer screening. So we'll play to the strengths of where we think the test is best positioned, and then we may look for alternative ways that may be in markets where low dose screening isn't as prevalent, we'll have market opportunities there. And then the last thing I want to clarify is that the market opportunity that we have shown here of roughly $4,000,000,000 between U. S. And EU, there is a much larger market opportunity for this exact test position where this one is, if you add up the China, Japan, Middle Eastern markets, Latin America, etcetera, etcetera, which we haven't done yet. And as we solidify our commercial plans and think about that more global opportunity, we'll bring those numbers forward as well. But we've started with U. S. And EU because we think that it will be important to stake our path for international expansion, and we believe this pulmonology franchise is the one to do it with. Okay. And actually, I do want to sneak one more here. Keith, can you just clarify the comments 2020? I don't know that I got that down, but you were talking about over 20% growth. I just want to make sure I understood what that comment was specifically referring and what it was excluding. So can you just reiterate that, please? Bonnie, she's talking about top line genomic and revenue growth of 20%, excluding biopharma. As you know, we have a lot of biopharmaceutical service revenue. We've been clear this year that we're not the J and J revenue will not. We have about $9,000,000 left on that to earn out of the 20 dollars And we will come forward in the Q4 call and talk about biopharmaceutical service revenue and where we think that's going to play out next year. But on product revenue what I would call product revenue, which is not it's still service revenue, but our classifier plus cyto revenue, she's talking about 20% top line genomic and revenue on those. Great. Thank you for that. Okay. You're welcome. All right. Thank you. Thank you. And our next question comes from Steve Unger with Needham. Your line is open. Hi, thanks. So you guys plan to commercialize in 2021 with the nasal swab test. That's quite a bit earlier than I think anybody expected. Is the expectation then to do an early access program similar to the other lung tests prior to Medicare reimbursement and then go nationwide? Well, as you might expect, I will probably say that as our commercial plans come together and we have clarity on that process and the timing of it, we'll certainly bring that all forward. We're under evaluation right now, and that's what will take us through 2020 to make sure we line all the pieces up to have an excellent execution on that launch. So I'm not trying to avoid the answer. I just think it's a little early yet to make any claim on exactly how we'll do it. I think we have lots of options. We're already in the Poems suite with these doctors. 80% of the Poems using Envisia are also using Percepta. So this will be one more way to add a test in the pulmonology suite and build the relationships with our customers. So that will certainly be part of the angle. But in terms of Medicare versus commercial and all of that, it would probably take us a couple more quarters before we have that information laid out. But thanks for the question. It's exciting. Got it. And then to expand internationally, particularly in the EU, are you planning the full pulmonology portfolio to expand that internationally or We have simply planted the seed right now that we do believe the global markets are big. It's very complicated, right, to tap some of the global markets. And we've always believed that we would look toward that planning once we had products and a portfolio that made more sense to make that effort on. And so we bring it up today as part of this data because we believe the nasal swab classifier could definitely be a pivotal part of thinking about that global expansion. We will be very thoughtful, I think as we always are, and we will give everyone plenty of notice on what that plan and timing and thinking will be. But there's certainly a big opportunity outside the U. S. To ignore when you have a test like this that can have such a great impact on care. Got it. And then if I could just one more. As far as the cost of the product, it's RNA Seq, right? It's on the same platform. What is the $800 at the low end of your is commercially viable given sort of the average cost per test that you're running at or could be running at in 2020, 2021, 2022? Yes. I mean, I think that we obviously will start where we can be in the upper end of that price range. We believe there is a good viable market for that. And we will enlighten you with other plans as those roll forward. But nothing different as far as costs relative to the Percepta, for example, as far as running the test and processing? Not right now, no. Got it. Great. Okay. Thank you. Ladies and gentlemen, this concludes our call today. Thank you for joining us. You may now disconnect. Everyone, have a wonderful day.