Burning Rock Biotech Ltd (BNR) 2020 Q3 法說會逐字稿

Ladies and gentlemen, thank you for standing by, and welcome to the Burning Rock's 2020 Q3 Earnings conference call. (Operator Instructions) Please be advised that today's conference is being recorded.

Before we begin, I would like to remind you that this conference call contains forward-looking statements within the meaning of Section 21E of the Securities Exchange Act of 1934 as amended and as defined in the U.S. Private Securities Litigation Reform Act of 1995. These forward-looking statements can be identified by terminology such as will, expects, anticipates, future, intends, plans, beliefs, estimates, target, confident and similar statements. Statements that are not historical facts, including statements about Burning Rock's beliefs and expectations are forward-looking statements. Such statements are based upon management's current expectations and current market and operating conditions and relate to events that involve known or unknown risks, uncertainties and other factors, all of which are difficult to predict and many of which are beyond Burning Rock's control.

Forward-looking statements involve risks, uncertainties and other factors that could cause actual results to differ materially from those contained in any such statements. Burning Rock does not undertake any obligation to update any forward-looking statements as a result of new information, future events or otherwise, except as required under applicable law.

With that, I would like to turn the call over to the company's Founder and CEO, Mr. Yusheng Han. Thank you. Please go ahead.

Thank you. This is Yusheng Han, the CEO and Founder of Burning Rock. So good morning, and good evening, everyone. Thanks for attending this call. And I'll start with a little bit of introduction of Burning Rock. We are Chinese molecular diagnostic leader for Precision Medicine. And there are 2 parts of our business. The first one is Therapy Selection, which means $4.5 billion TAM for China market. And the second one is Early Detection, which means $29 billion TAM for China market only. And our strength lies in 3 parts. The first one, we [have world-class technology] and the second is that we have a deep understanding and good experience about China regulations. And the third one is that we have a very strong network for hospitals and doctors.

So for Therapy Selection, our focus nowadays for are expanding leadership in China market based on our product performance and quality for both tissue and liquid-based NGS testing. Also, we enable higher NGS test penetration through kit-based in-hospital channel and that is complementing to existing central-lab based testing.

And for Early Detection, we are at the forefront of global R&D on a blood-based pan-cancer early detection. So in-house R&D and clinical collaboration with top physicians and top hospitals in China.

I'm very pleased introduce the speakers today, including me and our COO, Shannon Chuai; and our CTO, Joe Zhang; our CFO, Leo Li and our CMO, Chief Medical Officer, Hao Liu. And our topic today will cover 3 parts. The first part is for our progress for Early Detection. We will give everyone the product road map update, telling everybody that how we plan to do the clinical development for our products, no matter for the 6-cancer types or the 9 types of cancer tests. And I'm very exciting to say that we have a very good data just released at ESMO Asia Virtual Congress 2020. And we have 6-cancer test validation data release. That's very fresh and we will welcome to find a poster and read it. And our CTO, Joe, will explain that in detail later on.

And the second thing is about the milestones for Therapy Selection. And the first one is that we have our library prep automation system, Magnis BR, analytical validation data released at AMP, what we call Association for Molecular Pathology and a very, very good data out for both the tissue based and the liquid biopsy based. And the second news is that we just signed a licensing in agreement with Myriad for their myChoice HRD test. myChoice test as a gold-standard for PARP inhibitor. So we are very exciting and proud to say that, that test will bring a very meaningful test for all the patients of ovarian cancer and potential application for prostate cancer.

And the third part is our numbers. Our CFO, Leo Li, will explain that in detail, including the Q3 operation metrics and also the financials.

So let me transfer to our CTO, Joe Zhang, to talk about exciting Early Detection progress. Joe?

Thanks, Yusheng. So let's just get started on the Slide 7. So as a first, let's just recap the ELSA-seq, which is the technology behind our Early Detection program. So the ELSA-seq is a targeted deep methylation sequencing assay, utilizing the high-efficient library preparation as well as the machine learning aided classification model to perform the early detection. So the method can [query] methylation status from hundreds or thousands of the loci from the genome on the methylation status and from as low as 1 nanogram of a cell-free DNA to detect whether the ctDNA, what we call a circular tumor DNA is existing in the blood or not and also predict tumor of origin. So it has some kind of technology highlights being listed here. So including -- I just mentioned, there's a single-stranded library preparation as well as compatible with a different kind of conversion. And also, we select the most informative region from the genome for deep sequencing. Also, we applied the multiple noise reduction and the signal correction before our machine learning model building.

So the next slide talk about the Early Detection product development road map. So this is basically back in 2000, we started this program actually 4 years ago, but it's been under the research development mode. And right now, I mean, back to last year, in 2019, we presented the proof-of-concept of a single-cancer type, which is lung cancer in AACR Annual Meeting, and it's been listed in the first step. And in early this year, early 2020, we demonstrated 3-cancer performance in AACR Advance in Liquid Biopsy Special Meeting. And then we perform -- we demonstrated 95.1% specificity. At this specifically, we achieved 80 -- about 81% of sensitivity on 3-cancer type. As of like a couple hours ago, we actually just presented our 6-cancer model in ESMO Asia Virtual Congress as a mini oral session there. And in this -- the progress right now that we query about 6-cancer type, which are lung cancer, colon/rectum, liver cancer, ovarian cancer, pancreatic cancer and esophageal cancer. So we'll be able to achieve -- in the validations on cohort, we'll be able to achieve 98.3% specificity. And this specificity, we achieved 80.6% overall sensitivity. I'm going to talk about detail later in later slides. We also be able to achieve the tumor-of-origin results in the 98.6% of case. Among those, about like 81% are correct, which means we can predict 81% of the correct tumor-of-origing. For the next step, like currently, we are still develop further in terms of further expanded the cancer type, and we'll be able -- right now, the asset is still under development and the training and validation to follow. And in addition to the 6-cancer type, we added a couple more, including gastric and cholangio and head and neck. So this is basically the product development road map in terms of cancer Early Detection program happening in Burning Rock.

The first -- like, next slide that we just talk about the latest ESMO Asia result. And Dr. Gao Qiang from Fudan University, Shanghai Liver Institute had presented this latest data on multiple-cancer type early detection study. The study included 3 stages, including, what we call, marker discovery, panel design and validation as well as assay validation. So for marker discovery, we profile cancer and the non-cancer tissue sample in-house and also survey the TCGA and GEO public database to investigate a key methylation change associated with the 6-prespecified cancer type. As a result, we finally select about 160,000 CpG sites to construct -- customize capture-based panel to enable this cancer early detection. In asset validation stage, we randomly divide the plasma cfDNA sample from clinical diagnostic cancer patient as well as non-cancer control into a training set and also independent validation set. During the training phase, a machine learning-based classifier with a view to identify the presence of localization cancer derived signal. So subsequently, the model was applied to the validation site to evaluate the performance independently. As you can see, the overall work charts being listed in Slide 9.

And in the next slide, we basically listed overview of the training and validation set and how many patients -- individuals have been tested, including the cancer patient and a non-cancer control. The top table listed clinical characteristic of the training cohort and bottom-listed validation cohort. Both the case in the control groups are comparable with respect to age, gender and smoking status. As we know, my solution is very -- it's an age-related biomarker. That's why we've been really careful about this and make sure the age and gender and smoking status are being tracked and controlled. And about like 80% of the patients were diagnosed at early stage in this cohort, which means stage 1 to 3 at the study -- time of the study entry.

The next slide is a very critical slide, which is Slide 11. That's just the list of the summary of the detection accuracy or we can call it sensitivity across different cancer type and clinical stage. So we put 6 different box of bar graph here. The training set is labeled in green and a validation labeled in red or orange color. And the numbers in bracket and the number of samples included for the indicated stage and its group. In general, the performance was consistent in training and validation sets, and this has give us a lot of confidence and the detection rate increased along with the tumor progression from stage 1 to 4. In training, the specificity was being controlled at 99.5% and at this specificity sensitivity of 79.9% overall. And the in validation cohort, we will be able to achieve specificity of 98.3% with a sensitivity about 80.6%.

The next slide, basically recap of the number listed. This is Slide 12. Basically, we separate the number by cancer type and clinic stage and be doing a little bit of grouping here. And at given specificities, which are 99.5% for training and a 98.3% for validation, the categorized sensitivity are listed in the bar chart again. In validation cohort, especially at the orange to red color. Pay attention on that overall sensitivity from stage 1 to 3, we can call early sensitivity at 75.8%. In all stages, the liver cancer showed highest detection sensitivity of around 93% and the lung cancer was lowest around 67%.

The next slide, basically talking about the predicted tumor or tissue-of-origin with high accuracy. So to pay attention on the rightmost one, which is the validation cohort. So basically, we will be able to achieve around 80% -- actually it's from the 82% from the validation cohort, we'll be able to predict the correct tumor-of-origin, the tissue-of-origin. And actually, if you call top 2 organ call, which means like not only the top 1 organ call, we'll be able to achieve better. Actually, it's 87% from the validation cohort.

So for the next slide, basically, as Yusheng just mentioned here, we're going to talk about the Therapy Selection business. So we will have 2 topics. First, let me just give you some update on the Magnis BR, the walk-away a.k.a. turnkey solution for hybridization capture method in oncology panel testing. The instrument make actually current library prep method much easier and enable 9-hour from DNA to enrich the library for ready for sequencing in a fully automatic fashion. So this year, we actually made the Magnis BR now compatible to our most -- many -- several most popular panels including the liquid biopsy test, which is 168 genes panel as well as a 520 genes tissue-based testing. So this is being listed at Slide 15, just give you some kind of highlight in terms of the workflow and how the Magnis BR look like and why it's so useful for this kind of a turnkey solution for small-scale hospital lab.

Very exciting, basically, as a platform presentation selected by the AMP Annual Meeting. We actually presented -- be able to present this analytical validation result Magnis BR assay on this both assay, which is we call OncoScreen IO Tissue assay. This is a 520 genes assay as well as OncoCompass Target Cell-Free DNA assay, which is 168 liquid biopsy. And we did active validation on both panel, as we presented in the AMP Annual Meeting happened earlier this week. And the overall take-home message here is basically the Magnis BR assay showing equivalent or better performance compared to our manual assay and very high concordance being achieved and also the Magnis BR assay, it's actually -- we'll be able to fully take compatible with multiple different biomarkers, including the SME, in del, fusion and copy number variation as well as the microsatellite instability status.

Next, let me ask our Chief Operating Officer, Shannon, to give you more update on the Myriad myChoice product licensing activity. Shannon?

Yes. Thanks, Joe. So on Page #17, another exciting progress that Burning Rock in the last quarter is that we finally reached an agreement with Myriad to bring in myChoice HRD assay into China. So this assay is actually the first and arguably, the current gold standard for genomic instability calculation, which has become a very important and significant biomarker, especially for PARP inhibitor development. So far, the FDA has approved the myChoice HRD as a companion diagnostics for 2 PARP inhibitors, the olaparib and niraparib, both in the ovarian cancers. And we are listing the exact indications here on the table. So they were approved in 2019 and 2020, respectively.

And then on the next page, here, we want to introduce our thoughts behind the licensing in. So the purpose offering in this, I'd say, to China was, first, to improve our capabilities to collaborate with the pharmaceutical companies for biomarker, especially CDx biomarker development for PARP inhibitors in China because we see highly growing interest in that realm. And secondly, we also want to make this assay commercially available to Chinese patients in an LDP format at first. So our estimated immediately addressable market, which would be the first-line advanced ovarian cancer patients, who will undergo our maintenance therapy is around 10,000 to 15,000 patients annually. But I want to emphasize here that our estimated penetrants initially for these patients for that test would be around 5% in the first year or so. And then we anticipate it to grow, of course, in the long run.

And the second purpose was, of course, to host the long-run development for the -- and also a commercial move for this assay because we have a very optimistic view of the HRD as the biomarker for PARP inhibitors. And we are listing, I think that is a complete list of the PARP inhibitors and cancer types, which mychoice HRD is currently being used as a companion or exploratory biomarker in the clinical studies. So you can see that it has a great potential to be extended on the young ovarian cancer and also be on olaparib or niraparib in the future. So we are very optimistic in that, and we are also -- we want to sit in this good position to host such services for the pharmaceutical companies and also for hospitals in the future in China.

So next, I'm going to turn to Leo, our CFO, for the operating metrics. Leo?

Thank you, Shannon. And then turning to our numbers on Page 20. And first of all, let me recap that overall, we have had a good quarter during 3Q 2020, continuing our sequential recovery, growing in both the central lab and in-hospital channels on a sequential basis. And this is -- importantly, this is under the context of increasing competition in our industry in China. For example, we achieved 16% overall revenue growth on our top line on a sequential or quarter-over-quarter basis. And this is higher than the numbers we have seen from other public and private companies in our industry in China.

Looking specifically at our operating numbers on Slide 19. First, on the central-lab channel, we continued with the sequential recovery, growing our volume at 19% quarter-over-quarter. On a year-over-year basis, our central-lab volume grew 28%. Although there's no published industry-wide data, we believe we are gaining market share-based on anecdotal observations. For example, one company in our industry reported a 5% year-over-year LDP volume growth in the third quarter, which is lower than our 28% volume growth rate. Looking at the in-hospital channel, the third quarter has been a slow quarter in terms of new contracted hospitals. But on the other hand, the existing contracted hospitals continues their recovery and grew at good double-digit rate.

And let me discuss a little bit on the Magnis BR and recap that this was launched at the China Annual Pathology Conference back in November last year. The demand for Magnis BR beyond has been strong. However, we encountered some external supply issues in the previous couple of quarters. This has been gradually resolved, but it did mean that we have missed some demands, and we were not able to supply customers for a couple of quarters. At the end of the third quarter, we have placed Magnis BR into more than 10 hospitals in the low teens. But because of the delay in getting supply and the delay in procurements process, as a result we expect to complete sales contracts and afterwards book revenues for these instruments in 2021 for the majority of the low teens Magnis BRs that we have placed so far, with a minority spilling into 2022 due to hospitals's procurement cycles. The demand for Magnis BR is strong, and we expect additional placements into 2021. And Magnis BR as the only fully automated library prep system in China that can be large tissue and liquid panels is very important for us in terms of differentiating us from competition, and continue our leadership in the in-hospital market.

Moving on to our revenues on Page 20. Overall, our revenue grew 19% year-over-year. And the central-lab channel, which is more diversified geographically across China, grew 30%, and this is an improvement versus the 18% year-over-year growth rate in the second quarter. And the trend is in line with the volume trends that we just discussed, that we displaced on Page 19. The in-hospital revenue grew only 3% year-over-year and we'd like to explain the reasons behind that.

So first, we note that we had lumpiness in our quarterly in hospital revenues back in 2019 and this happens when we have a large hospital getting contracted in a given quarter. The lumpiness comes from one-off revenues getting booked in that quarter, the course of getting a contract signed for kits that were shipped in prior quarters before contract signing. So the third quarter 2019 was such a quarter and to a lesser extent, the first quarter in 2019 too. So you could see 2 clear lumpiness in those 2 quarters across 2019. We believe it's more representative to look at the trends across a number of quarters for the in-hospital channel due to the lumpiness in this channel. Numerically, if we were to take the average of 3Q '19 and 4Q '19 asset base, our 3Q '20 year-over-year in-hospital revenue growth rate would have been 41.6%. And sequentially, we should also note that the third quarter this year in-hospital revenue improved versus the second quarter and grew 15% sequentially. We had fewer number of newly contracted hospitals in the third quarter compared to the second quarter and this is an offsetting factor, dragging the growth rate lower in contrast to the continued recovery of the existing contracted hospitals. And we'd like to make a note that in the second quarter and third quarter 2020 all the in-hospital revenues were reagents' revenues. We booked revenue for one Magnis BR back in January this year, but because of the supply issues we just discussed, we were not able to supply or book revenues for additional Magnis since then. We expect to convert the currently placed Magnis instruments into revenues, but in 2021, and the demand for Magnis has been strong.

Pharma revenue drops in the third quarter, and this is due to reduced testing volume of Pharma projects in this quarter.

Looking for the ahead, we observed that COVID is still going around in China or lingering around with mini outbreaks from time-to-time and across a number of places. For example, we saw a few incidents of COVID cases, particularly in Northern China. And in a couple of those cases, these mini outbreaks led to population-wide testing for the entire city. And in one case, this led to the resignation of the Head of the City Public Health Administrator. So overall, the public health system and hospitals in China are still putting a lot of emphasis on COVID and it's -- and in this context, it is hard for us to make a precise prediction on how or when we will be back to normal. So we'll just have to continue to observe, and we will provide an update on our thoughts for 2021 at the time of our fourth quarter results.

And lastly, on OpEx, we expect increasing spend on Early Detection RMB going forward, as we are excited to continue to pursue this significant opportunity and increase our investments there.

So with that, we're happy to open up for questions.

(Operator Instructions) The first question comes from the line of Doug Schenkel of Cowen.

Just to start with a few questions on the asymptomatic cancer data presentation that you shared. First, just to start, I'm a little bit surprised that the way you cut this data, it seems like you toggled to higher specificity than what we talked about earlier this year. At that point, at least in our discussions and going back to the AACR Liquid Biopsy Meeting in Miami, you talked about maybe structuring the assay to have higher sensitivity at the expense of specificity, the normal toggle decision that one has to make in designing an assay and kind of going that way in a way that was really different than maybe what we see typically from companies focused on developing liquid biopsy assays for asymptomatic cancer screens in western markets. And I think you said that you were positioning it that way largely because of the differences in the screening test availability in China relative to the western markets. I'm just wondering why this changed in the way it's seemed to over the last several months?

Yes, this is Shannon. I can take that question, Joe...

Okay. Shannon, maybe, go ahead.

Yes. So first off, I want to make it clear that we didn't do it on purpose. So our philosophy stays the same. If we could sacrifice, if we could compromise specificity to gain largely on sensitivity, that would still be a route that we want to take. So moving from the 3-cancer assay to 6-cancer assay, actually we didn't just expand on the cancer types. There was a reshape of the marker selection and also the chemistry and also the algorithm. So this was not -- we didn't keep it at the same 95% specificity level on purpose, of course. But this is, again, looking at the -- this assay data separately and find the best kind of per se that we think is the good balance between sensitivity and specificity.

Okay. That's helpful. And maybe building off of that. The performance is impressive, especially when you look at it on a blended basis across all stages of cancers. And that said, when you pegged the results at a high specificity and then look at the sensitivity specifically for stage 1 lung cancer, which was 37.5%. I think it was 33% in stage 1 ovarian. It gets a bit better in esophageal and ovarian, where you get to 54% and 64%. And then, of course, liver and pancreatic is quite good in the [80s]. That said, for the first 4 cancers I mentioned, I'm just wondering if you think that's good enough for your target market. And I guess by extension, I'm wondering if your thinking is that given the absence of standard screening in China, at least in terms of availability, relative to western markets, that performance at that level would be good enough given the alternative is effectively finding no cancers. I'm just wondering how you're thinking about this because it does seem like the early stage detection for those cancers could be better, while keeping in mind that typically, these metrics actually deteriorate when you move the assays more into real-world studies?

Yes. I think you're absolutely right, Doug. This is not -- of course, this is not ideal. This is not like the ideal final ultimate performance what we want to see on early detection products, but we actually think this is already a very good performance product for Early Detection market, especially given that it's a pan-cancer or a multi-cancer assay. And keep in mind that when we say multi-cancer, we have to, also in the algorithm to, balance between sensitivities or the algorithms to balance the performance among different detects and also for the first, especially, it's highly correlated between different cancer types. So it's not like you can just adjust it, adjust the cutoff on one cancer type and not impacting the other one. So this is what our research team had so far figured out as the "back the balance." But of course, there are rooms for improvements possible in the future, especially, this is still relatively small cohorts. We are still learning a lot about the stage 1, stage 2 because actually, when we see in stage 1, stage 2, there are a lot of different a clinical features between -- even within stage 1, there are different clinical features and they each can have different performance. So in the future, when we have the larger cohort, we might be able to tweak it a little bit more precisely to see where can we gain and without losing too much on the other cancer types. And also I want to point out that this -- surprisingly, we find that now that we are in the 98 young specificities range, we actually still have a better sensitivity or a similar sensitivity at list with GRAIL, and it's showing. So we are also very interested to watch how people perceive the product performance on growth products. And we are also, of course, talking closely with the Chinese clinicians and see what their take is on that. And there we also have us to shape the -- to find the back balance probably when we do centralization. Joe...

Okay. Maybe -- I'm sorry. I was just going to add -- okay. Sorry to interrupt.

Doug, you mentioned China and just at one point -- so you mentioned China and just at 1 point that if we look across other Chinese innovators in this space, this data does look very strong. I mean we've seen more data around liver cancers in China. And this pan-cancer test results look strong even compared to others too, blood-based, single-cancer test for liver cancer.

Yes, absolutely. That data did look good. Yes. That's a good point. Just last on the studies. What comes next in terms of -- I guess what should we look for -- I guess, could you provide an update on PREDICT study enrollment and time lines or plans for any other large study?

Well, unfortunately, we can't provide any further details on PREDICT. We will -- as long as we have any update that we can release, we will share them with you guys, but not at this point.

Okay. No, that's okay. Can't blame me for trying. Leo, maybe just a couple for you, and then I'll get back in the queue. It was a better-than-expected revenue quarter. That said, we had expected the number of ordering physicians and number of hospitals ordering in the central lab to maybe increase a bit more than they did in the quarter. It did seem like you lost a little momentum there relative to what we saw in Q2. And it also seems like on a net basis, at least, that there were no additions to the in-hospital pipeline? I know you talked about some inherent lumpiness to some of these things in your prepared remarks, but I just want to make sure that these metrics weren't drastically different than what you expected. And if there was a divergence from expectation, I'm just hoping you could provide a little more detail.

Leo?

This is the operator. My friend, Leo, has just got disconnected. Allow me to call him back, and we will resume the call shortly. Please stand by. Thank you.

Thank you for your patience. Leo is now back. Maybe I'll invite Doug ask the questions again, please.

Leo, 2 of -- okay. Great. Actually, great.

A great question on the operating numbers. Apologies for the technical issue. So first, on the central lab operating metrics. And I would note that we are still on the COVID in China, so we were not able to get back to the peak breadth of reach or catchments that we achieved back in the fourth quarter '19. So that's -- I think that's still due to COVID being around China. We are, I think, doing well in terms of the number of patients tested per physician. So we are making progress there. But in terms of the breadthness of our coverage, that's with the COVID backdrop in mind. And the in-hospitals, I mean, this is one quarter and hospitals, they run annual cycles, so I think we should keep observing and rather than drawing too much of a conclusion out of a single quarter.

Okay. That's great, Leo. And very last one, and thank you for giving me this much airtime. Just in terms of competitive dynamics and pricing, I'm just wondering how impactful pricing changes from Illumina in China have been in terms of your positioning relative to others in the in-hospital market. We've heard from Illumina's public remarks and just other checks that there were some notable price cuts made in the Chinese market as well as in some other geographies with certain products, but specific to China. Have those helped you at all as you look ahead and think about your positioning in the in-hospital channel?

This is Yusheng. Are you asking -- you go ahead, Leo.

Go ahead, Yusheng.

So are you checking the questions about the impact of the in-hospital model.

Yes. Really just Illumina reducing the cost as your partner on the sequencing side in in-hospital market. I'm wondering if their price cuts have led to any improvement in your competitive positioning in China.

Well, that since most of the competitors are using the Illumina platform, they did cut price. And the good thing for us that we don't cut outside. So I think that, in the long run, means that we will have a better margin. But since you know that for most of the in-hospital model, sequencing cost is not a significant part for the overall cost. So we don't expect significantly lower -- I mean significantly higher margin, but will be a little bit improvement.

Next question comes from the line of David Lee of Bank of America.

Great. So basically, congratulations for your good third quarter results, especially during the pandemic period. You can still register a very strong growth. Basically, I have 2 questions. Number one is that recently we observed the press cut of centralized procurement policy are spreading not only on the medical -- not only on the drugs, but also on the medical devices. And it seems like there is a trend, and there is a chance that the PCR testing is also going to -- has the potential to be enlisted into the centralized procurement. Do you think that is something going to happen in near future? And if that is going to happen, how does this impact our business in the long run? So this is my first question.

My -- I've got another question. Is that -- can you share with us your understanding or the better understanding to the company -- to your 6-cancer sensitivity and specificity data because the sensitivity and the specificity are very strong. It's very high for the cancer in the Phase III and the Phase IV stage. But when it comes to the Phase I and the Phase II, it seems like some of the number is not that high. The previous analyst asked about your views on it. So this time, I want to ask in another way is that how does this data, especially the early cancer -- early-stage cancer data compared with your peers? And how does health authorities look at this data? Is that very strong? Is that a very high-quality something or specificity for the cancers at the early stage? Are we expecting these numbers to improve sometime later? So this is my second question.

Thank you. Okay. So for the first question, I would like to answer that. We think that it will be quite too early to talk about the -- the price cap from the government for IVD, especially for cancer combining technology products. The first reason is that for IVD product, usually the hospital takes a significant margin because we have a complicated operation in the hospital. So it's quite hard to claim the supplier is the reason to -- for relatively, I mean, for the price.

And the second thing is that the testing costs for the overall treatment for cancer, obviously, is not a major part.

And the third one, and it's a very important one, is that usually if there's something happened, similar thing happened for the IVD, it means that after we get the pricing, 3 years of the pricing, we will see the potential price cost or procurement from the government. So I would say that for our segment, it will be a long way, and we do not expect any change in the coming 3 years.

And in terms of the sensitivity and specificity of early detection, I will turn to our CTO, Joe, to explain about the data. Joe?

Yes. Just try to comment in terms of a different stage and different cancer type has different sensitivity. As you can see here, as just mentioned there, on stage 3 and 4 will be showing higher in general sensitivity. Just try to be aware, it's -- if you compare to our peer data, for example, GRAIL, and we currently have a very similar specificity, and we'd be able to achieve in terms of over 75% sensitivity for stage 1 to 3, which is a resectable, or what we call, removable cancer. And compared to their publication, I think we are still showing a little bit higher sensitivity than them, but of course this is still within the error range. We still -- we are currently are very -- we are optimistic, but we've been very conservatively optimistic on that. And we currently -- from R&D program perspective, we are currently and further refined our pipeline and the methodology and hopefully, by applying additional data featuring as well as the model building and trying to further improve the sensitivity, especially for the earlier stage for certain cancer type, for example, on lung cancer and ovarian cancer. And right now, it's still under development at this moment. I would just say we hope we will improve this early stage sensitivity, but I cannot tell you actually now how much we can approve -- improve it at this moment.

Yes. Right. Got you, very clear. That's helpful. Can I ask a follow-on question? So days ago, you announced the cooperation with Myriad. Can you share with us from the P&L perspective, how does this cooperation impact your potential revenue and potential profit? And what's your pricing strategy? Can you give us some color on this?

David, this is Leo, and I appreciate the question, and we are excited about commencing the partnership and hopefully, this will not be the only licensing partnership that we're going to have. So we are excited about leveraging our commercial infrastructure and bringing more tests to China for the benefit of Chinese patients. At this moment, we cannot comment specifically on pricing or revenues. We will need to go do a technology transfer initially before we can commence the test in China. So that will take a period of time, and we expect to launch this product at some point in 2021, and it will be at that time that we will be talking about the commercial aspects of this test. So I appreciate the patience there.

Next question comes from the line of Ethan Ding of Morgan Stanley.

And congratulations on a strong quarter. I want to ask 2 questions on the behalf of Sean. So the first one is still on the data. I kind of want to follow up on David's questions earlier. So I've seen that compared with your previous 3-cancer model, your 6-model data seemed some are higher, some are lower, maybe basically in line with your 3-cancer model. So how should we interpret this trend? And do you think this trend could be sustained for 9-cancer model? That's my first question.

And my second question is that -- so in this year's -- sorry, in this year's negotiation for the national reimbursement list more targeted cancer therapies may be included. Maybe in the next couple of years, more will be included as well. So how do you think that could drive the overall cancer genotyping markets, especially NGS, given that NGS is not reimbursable at this point? That's my 2 questions.

Shannon, maybe you can go ahead.

Yes. So for the 6-cancer type improvement over 3-cancer type improvement, I think we are optimistic about the trends that we will be maintain -- we will be able to maintain that performance when we expand into 9-cancer type or even more cancer type because we do have further ideas or activities to further improve our chemistry and algorithms once we move into the 9-cancer type assay. So there will be another technical improvement, which theoretically will bring a better performance, especially on the tissue-of-origin accuracy. I think there you will definitely see some room for improvement in the future.

For the early-stage cancer sensitivity, we are not sure we'll have to sit and see once we have more data come in, more data to train our model, as you know, this is a highly data-driven self-learning algorithm. So I think with more data coming in, we will, for sure, have a more robust performance. So the short answer is, yes, we are confident that we will be able to at least maintain and significantly improve in some dimensions, maybe not all.

Leo, so do you want to take the second question about the insurance?

Yes. On the second question, we do note increasing progress of volume-based procurements in therapeutics. And we believe -- we agree that it is helpful in terms of driving demands and increasing availability or affordability of targeted treatments for the benefit of Chinese patients, which will lead to increasing demands for patients with affordability issues. So it is a positive for the market overall. For us, it's less of an impact as we focus primarily on the top hospitals and centers, which have less of an issue on affordability. But overall, we think this is a good trend. And we look to complement the increasing affordability with increasing accessibility through our in-hospital model, by placing more tests into the pathology labs of hospitals, which is the typical way that patients are used to getting tested in China. So I think with the combination of increasing affordability and accessibility, these are helpful for increasing NGS penetration.

I want to add some additional comments for that. So I think for the top 200 hospitals, whether the NGS goes into reimbursement, it's not a key issue. If we have an investigation for the top hospitals with NGS have installed in the hospital, you can see that NGS now is taking a majority for the test for cancer patients. And we can see that PCR test is still increasing. The thing is that they are penetrating lower to more hospitals and that's a good thing. First, NGS testing is -- the penetration is just starting. And we see that PCR exploring the boundary of gene testing. And you can see overall the gene testing for cancer patients is increasing. So we don't see any increased situation for the whole market in the coming several years. Thank you.

(Operator Instructions) There are no more questions at this time. With that, ladies and gentlemen, that does conclude the conference for today. Thank you for your participation. You may now disconnect your lines.