Veracyte, Inc. (VCYT) Earnings Call Transcript & Summary

Good afternoon, ladies and gentlemen, and welcome to the Veracyte management discussion of the Percepta Nasal Swab Test pivotal clinical validation data. As a reminder, today's conference is being recorded. I would now like to turn the conference over to Tracy Morris, Veracyte's Vice President of Corporate Communications and Investor Relations. You may begin.

Thanks, Amanda. Good morning, everyone, and thanks for joining us today for a discussion of our Percepta Nasal Swab test pivotal clinical validation data, which we announced yesterday afternoon. With me today are Bonnie Anderson, Veracyte's Chairman and Chief Executive Officer; Dr. Giulia Kennedy, our Chief Scientific Officer and Chief Medical Officer; and Dr. Bill Bulman, our Medical Director for Lung Cancer. Before we begin, I'd like to remind you that various statements that we may make during this call will include forward-looking statements as defined under applicable securities laws. Forward-looking statements include those regarding our Percepta Nasal Swab test as well as 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. It also includes statements regarding the potential impacts to our business resulting from the COVID-19 pandemic and the potential timing for recovery of our business. 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 that 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 our May 19, 2021 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. Our announcement regarding the Percepta Nasal Swab test data is available on our website at veracyte.com under Press Releases in the Investor Relations section. We have also published a presentation regarding the data, which we will reference during our remarks. This presentation is also available on our website under Events and Presentations in the Investor Relations. I will now turn the call over to Bonnie Anderson, Veracyte's Chairman and CEO.

Thanks, Tracy, and thanks, everyone, for joining our call to discuss the pivotal clinical validation data of our novel noninvasive Percepta Nasal Swab test in lung cancer, which we will -- which will be presented at the upcoming ASCO Annual Meeting. To provide context, the abstract for the test was submitted in February and thus, focused on the test development. We now have, and ASCO has allowed us to present, pivotal clinical validation data for our final locked classifier. So the data has evolved from what you will see in the published abstract, and the new poster title is shown on Slide 3 of the presentation that Tracy referenced. The data we will be reviewing today are covered in the press release from yesterday and will be shared in a poster and audio presentation on June 4 during the virtual ASCO meeting. The poster authors include top thought leaders, including Dr. Carla Lamb from the Lahey Hospital & Medical Center, who spoke at our R&D Day in December as well as from other leading centers such as the University of Utah, and the poster will be presented by Dr. Peter Mazzone from the Cleveland Clinic, where he directs the Lung Cancer Program for the Respiratory Institute and the Lung Cancer Screening Program for the hospital system. I will start by providing an overview of the current diagnostic challenge in the early detection of lung cancer and how the nasal swab is positioned to bring much needed clarity for physicians and patients as part of a broader portfolio of Veracyte tests for lung cancer. Then, as shown on Slide 4, we will feature Dr. Giulia Kennedy, who will provide an overview of the validation data that will be presented in more detail at ASCO. Dr. Bill Bulman, who recently joined Veracyte as Medical Director for Lung Cancer, will verify context on the impact we anticipate the nasal swab test will have on physician decision-making and clinical practice. After our prepared remarks, we will open the call for your questions. So turning now to Slide 5. Lung cancer is the biggest cancer killer worldwide, largely because most cases are not diagnosed until they are at an advanced difficult-to-treat stage. Thus, early detection of lung cancer is key to saving lives. Lung cancer nodules are typically the first sign of lung cancer, and they cannot be ignored. However, most of them are benign. Currently, we estimate about 1 million patients with suspicious lung nodules are sent to a pulmonologist for a workup each year, with most nodules found incidentally. Lung cancer screening programs are increasing, and I remind you of the new U.S. Preventative Services Task Force recommendation, which expands the number of people eligible for annual lung cancer CT screening to an estimated 15 million. So we expect the number of patients identified with suspicious nodules will continue to grow. Unfortunately, when evaluating these nodules, physicians currently lack standard and objective tools to determine the likelihood that they are cancerous. This uncertainty can lead to unnecessary invasive diagnostic procedures or potentially to delayed diagnosis and treatment. Enter the Percepta Nasal Swab. As illustrated on Slide 6, the first of its kind test is designed to provide physicians with objective clarity to better guide patient care. As outlined in our press release yesterday afternoon, the pivotal clinical validation data suggests that our noninvasive test can significantly improve early lung cancer detection. In simple terms, the nearly half of patients with suspicious nodules, we believe the test will allow physicians to accurately and confidently direct patients identified as low risk to routine monitoring and direct those identified as high risk to further diagnosis and potential treatment. Based on our discussions with thought-leading physicians, we believe this novel test will be transformative for patient care, potentially helping patients with benign nodules avoid unnecessary and costly invasive procedures and enabling more lives to be saved. We are on track to launch the Percepta Nasal Swab test in the second half of 2021, at which point, we will make it available to a select number of sites through our early adopter program as we seed the market and amass the clinical utility evidence needed to obtain Medicare and commercial payer coverage. As you can see on Slide 7, we have a comprehensive lung cancer franchise. We plan to adapt the nasal swab test, the nCounter Analysis System, in 2022, which we believe will enable us to begin making the test available to patients in global markets in 2023 and thus access an approximately $8 billion global market for our lung cancer franchise. Development and validation of the Percepta Nasal Swab also marks a key milestone in our ongoing collaboration with the Lung Cancer Initiative at Johnson & Johnson. I will now turn the call over to Giulia with a brief introduction, as shown on Slide 8. Giulia has been with Veracyte since our inception and is the scientific visionary behind all of our legacy tests including Afirma, which has become a standard of care, transforming how thyroid cancer is diagnosed and saving more than an estimated 100,000 patients from unnecessary thyroid surgery. It is Giulia's insight from the beginning on using a whole-transcriptome approach to our tests that have enabled us to develop such high-performing tests and have positioned us so well to expand our offering along the care continuum and forge powerful collaborations with leading biopharmaceutical companies.

Thanks so much, Bonnie. Good morning, everyone. Let's turn to the development of the Percepta Nasal Swab on Slide 9. Our nasal swab classifier was developed from a rich training set of nasal samples from over 1,100 patients. The nasal swab test uses an established field of injury principle to detect smoking-related damage in the airway, which is associated with lung cancer and can be detected in the nose. We use machine learning on whole-transcriptome RNA sequencing data from nasal swabs, along with clinical factors, to develop the algorithm, which is comprised of 502 genes. Two decision boundaries were chosen to maximize sensitivity and specificity for low and high-risk nodules, respectively. The rationale behind this 2-cutoff approach is that we set up the training of the classifier to distinguish between confirmed benign and confirm malignant samples. Our models did a really good job of separating many of the truly benign samples and many of the truly malignant samples from each other at the lower and upper ends of the score space. As a result, we could employ 2 cutoffs to take advantage of this separation, meaning, that 1 cutoff ensured a nearly pure population of benign patients that could be designated as low-risk, and the second upper cutoff ensured a population enriched in malignant samples that could be called high risk. Samples that are not classified as either low or high risk are classified as intermediate risk. This classification of each patient into 1 of 3 risk categories, low, intermediate and high risk, provides results that align with the same risk categories used in the American College of Chest Physicians, or ACCP, guidelines. We believe this harmonized approach will facilitate the test's eventual incorporation into these guidelines. We clinically validated the lock classifier on an independent test out of 249 patients. The test set was comprised of multiple cohorts of prospectively collected nasal samples of current or former smokers undergoing evaluation for lung nodules found on computed tomography, or CT. This test set comprise 2 important data sets, the AEGIS and Lahey cohorts from a total of 25 sites, of which 47% were from academic centers, 43% were from community settings and 10% were from the VA system, helping to ensure diversity. All patients were followed for up to 1 year or until a final adjudicated diagnosis was made as either benign or malignant. The cancer prevalence in the validation cohort is 54%. Since we believe our test population will have a lower prevalence of cancer, approximately 25%. I will show you the validation results at both the study prevalence of 54% as well as at an extrapolated prevalence of 25%. As you can see on Slide 10, results from the clinical validation study shows that when the test identify patients as low risk for cancer, its sensitivity was 96.3%, specificity, 41.7%. Note that for ruling out cancer, high sensitivity is the key factor. When the test identified patients as high risk for cancer, its specificity was 90.4%, sensitivity, 58.2%. Note that high specificity is important when you're trying to rule in cancer. These findings show that the test classifies more than 40% of patients with truly benign nodules as low risk, allowing them to avoid further procedures. The test also classifies nearly 60% of patients with true malignant nodules as high risk, enabling them to be directed to more timely diagnosis and potential treatment. The remaining patients were classified as intermediate risk for cancer, and these patients would be managed according to the current standard of care. As you likely know, sensitivity and specificity do not change with prevalence of cancer, but negative predictive value, or NPV, and positive predictive value, or PPV, do. First, let's look at the NPV and PPV at the study prevalence of 54%. We show that at this high cancer prevalence, the NPV of the nasal swab test is 90.6%, which is very exciting, given how hard it is to reach a greater than 90% NPV at such a high cancer prevalence. This is because our sensitivity of 96% is very high. The PPV of the test at the study prevalence is 87.8%, also very high. We are very pleased with these validation results. When we look at the NPV and PPV at an extrapolated 25% cancer prevalence, which is the level we expect in the broad population of patients with suspicious nodules on which the test is expected to be used, we also see very exciting results. For this population, the NPV is extremely high at 97.1%, meaning that a patient classified as low risk has only a 2.9% risk of malignancy. Similarly, the test had a positive predictive value, or PPV, of 67% and when applied to a population with 25% cancer incidents. This means that nearly 70% of patients classified as high risk will have lung cancer. Current ACCP guidelines recommend biopsies for patients with greater than 65% lung cancer risk, again, underscoring that our risk categories are aligned with guidelines. We are very pleased that our test performs well at both at 54% and 25% cancer prevalence. This is not an easy accomplishment. These results should give physicians significant reassurance in telling the patient deemed low risk by the nasal swab test that they have a very low risk of cancer, regardless of whether that patient came from a screening program or because of a nodule found incidentally. These findings suggest that the nasal swab test will be able to objectively and accurately stratify nearly half of the patients with lung nodules found on CT scans to either low or high risk, while those classified as intermediate will remain a candidate for current standard of care. We're excited about the opportunity to transform the early assessment of lung nodules with a simple nasal swab test. I'll now turn the call over to Bill to put the performance of the classifier into perspective of its anticipated real world use.

Thank you, Giulia, and good morning, everyone. I am very excited to be joining the call today to discuss the validation data for our nasal swab test. This is truly exciting data on a first of its kind genomic test for lung nodule risk assessment, an area where I've personally seen the life-saving impact of early detection and the devastating consequences of late-stage lung cancer. There is a definite opportunity here to improve the standard of care and to potentially save more lives while at the same time reducing costs for the health care system. Before I get into this further, turning to Slide 11. I just want to briefly express why I jumped at the opportunity to join the Veracyte team. I've been a physician for 27 years and a pulmonologist for the last 15 years. Until just a few months ago, I served as the Director of the Lung Nodule Assessment Program and the Director of Bronchoscopy at Columbia University Medical Center in New York. I've been an active user of Veracyte's Percepta GSC product, and I was an investigator in the AEGIS study, which was the early origin of the nasal classifier that we are sharing with you today. When the Veracyte medical team reached out to me, and they shared the data on the coming nasal swab test and the data on the soon to follow Percepta Genomic Atlas, both of which I believe are going to be paradigm-shifting additions to the management of patients with lung nodules and lung cancer, I just jumped at the chance to join their team and help move this forward. Turning back to the risk assessment of a lung nodule. From the physician's perspective, I'm excited that the nasal swab test will soon be available to help guide patient care using the combination of genomic science and machine learning. Currently, physicians lack reliable data-driven tools to determine which patients with a lung nodule found on CT scans have cancer and which do not. To provide some perspective on the tools that we currently do have, the methods available to clinicians now include risk calculation models, which are based on a limited number of clinical factors, or evaluations based on a clinician's experience, the latter is referred to as physician assessed risk. There are great challenges with both of these methods. The risk prediction models are validated based on the population from which they were derived, which may lead to over or underestimating risk for a patient who is not represented in that original cohort. And physician assessed risk is highly subjective and highly dependent on the physician's experience. The models and physician assessed risk are highly influenced by size, so they'll therefore see small cancers and things that are benign, and they'll see larger lesions that are benign and overestimate the risk of cancer. The whole process of assessment of risk is incredibly stressful, even for a pulmonologist with 15 years of experience, but it is critical for determining what is the most appropriate next step for a particular patient. Turning to Slide 12. You will see a slice from a CT scan, which shows a 15-millimeter lung nodule discovered in a 68-year old man who smoked for 35 years. It is frightening, okay? It is a difficult module to biopsy with a bronchoscopy, and it is a very risky nodule to biopsy by other means. And surgically removing this nodule would require removing half of the entire lung on that side. Decision-making is hard here. Should I try to biopsy this? Do I follow this over time and potentially just lay a diagnosis for a patient with cancer? Do I send this person to surgery for what might turn out to be a benign nodule? From the physician's perspective, the nasal swab's test ability to provide confidence knowing in who to direct to surgery, who to refer for a biopsy and who to reassure, telling them that they, in the context of a low-risk result, likely do not have cancer. All of this will have enormous impact on lung module management. I also think it's helpful to think about this from the patient perspective. Leads this individual patients -- and imagine a patient who has smoked for 20 years and quit 10 years ago. This patient will be told by their doctor that they're high risk for lung cancer and need to be screened. And imagine on their first screen, a small 7-millimeter nodule is identified, less than half the size of our patient here. Based on this information, because most small modules are low risk, the radiologists will classify it as a benign nodule and direct the patient to CT surveillance. This patient will have to wait 6 months to find out if the nodule is, in fact, benign or if it has grown and is cancerous. If that patient truly has a noncancerous nodule, which they likely do, why should they have to wait 6 months for confirmation of that? If this person truly has cancer, why should they have to wait 6 months for confirmation of that diagnosis and for curative therapy? With the nasal swab test, we have a new tool that can accurately identify the personalized risk for this patient based on their completely unique genomic analysis and give the patient some comfort that they can move forward either to surveillance or confidence that they need a more aggressive approach. I view this as transformative not only for the potential to save lives and reduce costs, but also for the patient journey, both for patients with cancer and patients without cancer. Leaving that hypothetical patient and getting back to the real run in front of you. The nodule that you're looking at, the 68-year old man with 35 years of smoking history, turned out not to be a cancer. The Gould model, one of the risk calculators, gave this person a 62% risk of cancer. It took nearly 4 months and a nondiagnostic procedure to determine that they did not have cancer. I believe -- we, at Veracyte, believe that the nasal swab test will bring objective certainty to lung nodule evaluation, delivering the best care decisions for patients. Our findings show that it accurately determines which patients are at low risk of cancer, which are truly intermediate or moderate risk, they need to have a biopsy, and which are at high risk and should get an aggressive approach. We believe that providing a simple, noninvasive test with all of these capabilities, particularly one that misses very few cancers, a test done right there in the doctor's office at the initial evaluation, could be a game changer for early lung cancer assessment. And as Bonnie mentioned, this is very good news for the 15 million patients who are recommended for annual CT screening on the basis of being at high risk for cancer as well as the 1.6 million patients a year with lung nodules found incidentally on CT scans. It has really been my pleasure to talk with you today. And with that, I turn things back over to Bonnie.

Thank you, Bill. We'll close with Slide 13 and the fact that we are extremely excited about this data and the potential of the Percepta Nasal Swab test. We believe it's going to be a game changer in early lung cancer detection. It will also be a lynchpin for our comprehensive lung cancer testing portfolio, which includes our Percepta Genomic Sequencing Classifier to improve lung cancer diagnosis and our in-development Percepta Genomic Atlas, which will detect gene alterations that may inform treatment decisions from the same small biopsy used for diagnosis. Collectively, we believe our test will provide important answers and insights that enable physicians and patients to make better, faster and more confident care decisions at every step of the patient's journey. With that, I'd like to open the call up for questions.

[Operator Instructions] Your first question comes from Brian Weinstein with William Blair.

I guess I'll just start with as it relates to physician assessed risk, which you were just talking about Dr., versus this product in particular. I'm curious, in your experience, do you think that a product like this is sufficient enough to have clinicians kind of override their internal suspicions here when telling them that they can down classify a patient to, let's say, intermediate risk or something like that as opposed to being high risk? And I'm wondering what kind of clinical utility data that you guys think that you'll need to show to kind of validate that the test is actually being utilized in the way that it's meant to be by the clinical community. So we'll just start with that one.

No, that's a great question. Thank you very much for that. This is not a test designed to override physician assessed risk in any way. It's designed to be a data-driven piece of analysis that adds to the physician risk assessment. Physicians are going to put all of the data that's in front of them and the patient's unique clinical characteristics into context, and this provides some objective data that will allow them to formulate the most appropriate next plan. One of the major problems with lung module management is that physicians don't really follow the guidelines. They, in certain circumstances, underestimate risk and overestimate risk and sometimes they are very gun-shy about taking patients and even patients that they feel are high risk for cancer and taking them -- sending them to a thoracic surgeon to have the nodule removed. This is data that allows a high-risk patient to be shown some evidence that they truly need aggressive therapy. And on the other side, many, many -- most nodules coming through a nodule assessment program are going to be benign nodules, and those patients are terrified that they have cancer. Even if the physician thinks that they're low risk for cancer, convincing them with a piece of data is going to have enormous impacts. Look at this assessment of your own individual unique genomic signature, it says that you are with -- in the context of this nodule, unlikely to have cancer, and we're 97% confident. It's hugely impactful. So you have impact at both ends. That's the kind of data that we are going to need to see in a clinical utility study across the risk spectrum of all nodules, nodules that are likely to be benign, nodules that are likely to be malignant that physicians use the nasal classifier to follow guidelines and to reassure patients. So we're going to be looking at adherence to guidelines, we're going to be looking at outcomes, and we're going to be looking at the impact on patient levels of anxiety related to their plan, whether it's for a benign nodule or for a cancerous nodule.

Thanks, Bill. Yes. Brian, Giulia also wanted to speak to some of these calculators.

I just wanted to add to Bill's assessment, is that there's a lot of data out there that show the current risk calculators. These clinical models out there actually performed pretty poorly in clinical practice, and many physicians, as Bill alluded to, lack confidence in using them or don't use them at all. And since they're driven a lot by size, they see small cancers as benign or intermediate, they call all large infectious nodules cancer. This same failing can happen to the physician gestalt, if you will, their determinations of risk. So small cancers may be dismissed as inconsequential, leading to delay in diagnosis and treatment. So we're just finishing an analysis of patients with lung nodules who are deemed to have enough of a risk of cancer to justify bronchoscopy biopsy. The bronchoscopy's ability to predict the cancer risk was actually quite poor, and the risk calculators were no better and sometimes worse than a coin toss. So we do expect to submit that study for publication in the coming months. And I think that will shed a lot of data-driven evidence about how these clinical classifier actually do on the real world.

And I want to follow-up on that because that was actually leading into my next question on things like Mayo and Gould and all these other risk calculators that are out there. With respect to your product that you have now, how much of the improvement is making adjustments off of more clinical factors that were out there that you could improve those kind of more standard risk calculators that are out there versus the 500-plus genomic markers that you guys are bringing forward? I suspect that almost all of it is in the latter, but I just wanted to confirm kind of how you break down the impact from the transcriptome analysis and how much of the benefit that that's actually providing. I assume it's almost all of it, but I just want to verify it.

Yes. Thanks, Brian. Yes, in fact, we actually make use of clinical factors. We interact them with the genomic data in cases -- in some cases. And so we actually make use of both. But the point is that we're doing it in a very data-driven way with labels of benign and malignant to guide us. And so I think the proof is in the pudding, if you will. The validation is showing really great performance of the classifier. And we also show that it does a better job of clinical factors, the Gould model, in doing a low-risk classification. And more of that data will be forthcoming over the next couple of months as we put together the publication and submit that for a public peer review.

And lastly, Brian, you mentioned clinical utility. We know how to do clinical utility. We're looking forward to getting the product onto the market. We've got lots and lots of sites lined up for early adoption, probably more than we'll be able to bring into the program to begin. And we will build the confirmatory evidence that this test indeed can really change care for patients with better outcomes and a lower cost of care. So we look forward to that. That will be our early step after launch.

Your next question comes from Sung Ji Nam with BTIG.

Congratulations on the data. Just kind of curious, maybe Dr. Bulman, given you're a user, until very recently, of Percepta GSC, was wondering kind of how are you looking at, given the availability of the nasal swab as far as kind of evaluating your patients for potential further workup or bronchoscopy, et cetera, how are you kind of looking at these products collectively because there are a lot of synergies here. Would love to get your thoughts on that.

There's definitely -- that's a great question. Thank you. Yes. Thank you. No, there's definitely synergies here. The nasal swab classifier is an initial assessment done at the time that the novel has been identified. Percepta GSC was developed to be a risk reclassifier in the event of a nondiagnostic bronchoscopy. So those patients already have been deemed to be -- the Percepta GSC patients have been deemed by some risk assessment up until now, it's been the physician's best risk or the lung calculators, some risk assessment that has determined the patients of sufficient risk to undergo a biopsy, and that's generally the intermediate category. It's the low-to-moderate category in the ACCP guideline. Once that patient is undergoing a biopsy bronchoscopically, they have about 40% to 50% chance of having a nondiagnostic result, depending on the modality used to [ bronch ] that person. Even some of the best electromagnetic navigational tools to guide bronchoscopy, the yield is 70%. So there's going to always be a significant number of patients who have a biopsy, and we get a nondiagnostic result. And the GSC will continue to be sort of insurance for that nondiagnostic biopsy. It's another piece of data that helps modify the risk. And because GSC was developed in this intermediate risk group of patients and developed specifically for patients who are undergoing bronchoscopy, it's not exactly the same classifier as the nasal classifier. It was trained differently and validated differently. So we see it having a role sort of in succession with the nasal swab classifier a patient will get deemed by the nasal swab classifier being of intermediate risk, sufficient to justify a bronchoscopy but not sufficient to risk, not sufficient to justify going right to the operating room. And then GSC will be done at the time of the bronchoscopy to give the patient data in the event of that nondiagnostic bronchoscopy. Here, we have additional genomic signal that tells us -- in fact, despite the intermediate classifier, you're actually, on our GSC classifier, now bumped up to high risk and it's time to be aggressive and move on to therapy. And likewise, if they're down classified by GSC, the aggressiveness of the approach may be dialed back. So we definitely see synergy between these 2 products. Nasal is not a replacement for GSC in any way.

Great. And then I have just one more follow-up, sorry. But did you -- sorry if I missed it. But did you -- what's the turnaround time for the nasal swab test?

Our turnaround time will be guaranteed 5 to 7 days typically, like the rest of our tests are. And that's actually plenty of time. When the patient's determined to have a nodule, we can take the nasal swab test. And by the time the patient comes back for evaluation and next steps, they'll have all the results to make the best decision for the patient.

We also have vision of potentially, if this is identified as a screening nodule, having the screening, nurse care coordinator swab the patient at the time of the CAT scan. Most CT screens are read in real time. And if there's a worrisome nodule, they could get a referral and a swab. And so that by the time see the pulmonologist for their initial evaluation, they're having discussions -- data-driven discussions right then and there. There's definitely potential to use the turnaround time to advantage patients.

Lots of great opportunities here to integrate into both the current standards of care and create new standards of care as we bring these tests along.

And your next question comes from Tejas Savant with Morgan Stanley.

Congrats on the data here. I had a sort of big picture theoretical question actually for Giulia, if I may. On this cohort creation that you spoke about, moving to that 25% estimated prevalence. Can you just walk us through how exactly you did that? I mean is it essentially sort of dropping out some of the cancer incidence from the 249-patient cohort? And then why does the NPV come down meaningfully in the high-risk patients as you move to the extrapolated cohort?

Well, first of all, it's extrapolated. So that's a mathematical transformation of the data. So this is why we presented both the 54% prevalence, which is the prevalence in the study, that affects the NPV and PPV. It does not affect the sensitivity and specificity and the 25% prevalence, it's just math.

Also, I want to just mention. On Slide 10, we had a typo on that slide, where we are referring with a very end data point, 67%, says NPV, that's actually PPV. It's been fixed now. Okay, good. Thank you.

Got it. That's actually -- that's helpful.

G So just focus on [ sensitivity ] and specificity. And the NPV and PPV at both prevalences shows we're thrilled that the data looks so good at both that very high prevalence of the study as well as when you extrapolate it to a 25% incidence of malignancy. It's we're excited.

Yes. Can't get much better than that.

Got it. Got it. And Bonnie, can you comment on just performance for the test by nodule size or length of smoking history? And then as a follow-up to that, I know you mentioned earlier -- at least the abstract mentioned sort of a focus on the current and prior smoker population. So in terms of the incidentally detected patients here, what fraction of those patients are typically non-smokers in your experience?

Yes. There is a definite percent of patients with incidental nodules that don't come through the funnel with smoking as a history. Fortunately, everyone that is being entered into the screening programs are and that's where this is going to really help get more and more patients that need to be screened into the screening program and be that noninvasive next test. As the data get further evaluated, we'll come out with subset analysis that will include nodule size, stage of lung cancer and all of that out of the data set. That just isn't part of what's going to be presented in the first set of data at ASCO. So that gives everyone lots of great data to look forward to.

Got it. [indiscernible]

The incidental lung nodules that are being referred to lung nodule assessment centers, like the one I had at Columbia, they're very enriched for smokers because those are the people that are -- the patients that are terrified that they're 8-millimeter finding on a CAT scan could be a lung cancer, and that's a valid fear. The vast majority of cancers are driven by current and former smoking and the risk associated -- the DNA damage associated with that, so those patients are more likely to be referred for a nodule evaluation if they have a history of smoking with an incidental nodule.

Thank you, Bill. That's right.

And Bonnie, thank you.

You can ask -- I'm sorry, Tejas.

All right. All right. No worries. Just one final one for me. I just wanted to ask is the classifier essentially locked down now, given that you'd be using this validation data as the basis for submitting to payers, et cetera?

Absolutely, [ locked ].

Or is there room for further improvement? Or are you done?

It is locked. There's always room for further improvement. We call that version 2 down the road. We've done that with just about every one of our classifiers. It's fully locked and ready to go.

Your next question comes from Thomas Flaten with Lake Street Capital.

Just a quick one, Bonnie. I don't know if you can even give us maybe a range of how many sites you anticipate launching this into as we come towards the end of the year.

Yes. We will probably go to somewhere between 50 and 100 sites for the early adoption program. We like to keep it somewhat limited so that we can really collect great data and bring through the publication process good clinical utility data and make sure the test gets seated in these early sites as a real standard of care operationally. So we've got lots of names on the list already. And then over time, as we achieve our Medicare coverage decision and commercial coverage decisions, that's when we will break out of that and start to ramp up the number of sites. We should be able to collect a pretty good amount of data early on because the nodule volume here is much larger than it typically is in our other tests. So we could get a lot of great clinical utility data from just 50 sites. Thanks for the question.

Many of these sites are also going to be places where the pulmonologists who write the guidelines practice. And so they'll get to use this in clinical practice and see how it improves patient care. And hopefully, given our great alignment with the guidelines and the potential for it to increase physicians' adherence to the guidelines on the basis of having objective data, our hope is that we can be included in the next version of the ACCP guidelines.

And then just to confirm. This will be launched as a sequencing test, right, not any other modality [indiscernible].

That's correct. Yes. They'll still come out of our unified assay [ CLIA lab ] here. And then as we -- as I mentioned in the prepared remarks, we also have plans to adapt this classifier to our nCounter so that we can make it available to patients all over the world. This is going to be an exciting journey.

[Operator Instructions] Your next question comes from Mike Matson with Needham.

Yes. So I understand the commentary about how you're viewing Percepta GSC as kind of complementary and it fits differently into the tech care pathway. But at least according to your website, it has 91% negative predictive value and 65% positive predictive value. So -- I mean, which actually seems a little lower than what we're seeing here with this -- the nasal swab test. Is that not an apples-to-apples comparison? So how do we -- I guess, how do we view this data in the context of the Percepta GSC data?

Yes. Well, it is not apples-to-apples comparison. It's a different test. Giulia, do you want to describe?

Yes. So the Percepta GSC, it is on bronchial brushes taken from the airways. So it's a different sample type. It's a different classifier. It is also based on sequencing data and some clinical factors, but it's a different algorithm. And so it's going to have its own performance characteristics because it's a different set of patients. These are patients that undergo bronchoscopy. The nasal swab classifier is developed on swabs taken from the nose and also sequence, the data will be different, whole transcriptome, clinical factors, but the models are different. And so they're each held to their own set of product requirements, if you will, and develop to meet those specs. So you can't really compare the numbers.

Okay. That's helpful. And then if this was done in a population -- if the study has been done in a population with 25% cancer rates instead of 54%, would that imply that around 30% of the patients would be low risk, 55% intermediate risk and 15% high risk? In other words, I'm just taking the percentages that you found in those categories, in the 54%, multiplying it by the number of patients with cancer and without cancer in the lower prevalence population.

I mean the way we like to think about it is regardless of prevalence of the people that are truly benign, 41.7% of them will be called low risk, and then of the people who are truly malignant, 67% will be called high risk. So that kind of gives you an idea, regardless of prevalence, what's happening, which is a truly benign or truly malignant population. That's how that classification will work.

Okay. But around -- but I think you said early around half the patients [indiscernible] is going to be in a category...

Yes, around half will be -- that's right. About half will be classified to high risk or classified to low risk. And the intermediate risk group will remain in that intermediate risk category.

I misspoke there for a minute. I'm sorry. I -- 58% of the truly malignant patients will be classified as high risk. I was looking at the wrong number.

And that intermediate risk category is the bread and butter of our -- for our end users, the pulmonologists, the -- deal with the intermediate category by referring them for a bronchoscopic biopsy generally or for a transthoracic needle biopsy. There is a well-established guideline driven plan for those patients deemed to be of intermediate risk.

Okay. And if I could squeeze in one more. If the patients end up in the high-risk category, would they still get a biopsy? Or would they go straight to some surgery or treatment for cancer?

Generally, the guidelines for the last 9 years have dictated that patients who need a certain threshold of risk should not get an intervening biopsy. They should go right to some form of definitive therapy, either surgery or SBRT or radiation therapy. The reason being that the biopsy has morbidity associated with it. And if a patient has a very high risk of cancer and you prove that with a biopsy and then take them to the operating room, essentially, you've had an unnecessary step there, an unnecessary procedure. And likewise, we always worry about -- pulmonologists always worry about the false negative. You biopsy somebody that looks like they have an intermediate risk of cancer, and you get back a nondiagnostic result. And you cannot be falsely reassured that the wrong thing has been determined by this biopsy, the wrong decision's been made. So the fear of a false negative drives people to the operating room also. So if everyone's going to the operating room, irrespective of whether they have a positive biopsy or a negative biopsy, then the biopsy becomes really unnecessary. That's not for all patients. There are some patients where we have the biopsy because surgery is too risky, or we have to biopsy for other reasons to convince the patient that they need an operation because otherwise, they don't agree to the plan. But the pathway to surgery for somebody who is high risk is established. We are staying with a high level of confidence based on a high-risk result that a patient has a malignant lesion and should follow that pathway.

Thanks, everyone, for joining us today. I think as you can tell just from the questions and the conversations, this is a very ambiguous area for patients -- and for physicians and patients to navigate. We are truly excited that we are going to be able to bring the first of its kind noninvasive nasal swab to be able to guide next steps for patients with high degree of accuracy and certainty, and we're excited. So thank you all for joining us today.

That does conclude today's call. We thank you for your participation. You may now disconnect.