Perfect. I think we can get started. Kallum Titchmarsh here from the Life Science Tools and Diagnostics team at Morgan Stanley. Thank you all for joining us today, joined by Eric Fung, Chief Medical Officer at GRAIL, for a fireside chat discussion. Just before we get started, for all disclosures relating to this presentation, please see morganstanley.com/researchdisclosures. Eric, maybe for those that haven't had the opportunity to meet you, could you just tell us a little bit more about yourself and your involvement with GRAIL during those nine years I think you've been there?
Yeah. Well, thank you, Kallum, and thanks for the Morgan Stanley team for having us here. I'm Eric Fung, Chief Scientific Officer for GRAIL. I've been at GRAIL, as Kallum said, for nine years. I've really spent my life, my career, developing novel diagnostic tests, and when GRAIL was first announced in 2016, it was really fascinating to me, this idea that you could screen for cancer using just a blood draw. You may recall at that time, though, it was about the time that Theranos was really very much in the news. Talking to my colleagues, there was a lot of, "Can they really do that? Is it real?" I decided to take the leap, and I've been working at GRAIL ever since to really develop the clinical evidence program, really bring the science to bear, and it's been really paradigm-changing.
Perfect. GRAIL's obviously spent a lot of time and dollars throughout that period building the evidence and infrastructure behind Galleri. Where do you think the company is today on the path from a kind of niche early adopter product to something that could become more routine in that preventative care?
There's several steps along the way. As you point out, early adopters are really important, but obviously clinical evidence is the single key driver. That clinical evidence opens many doors, and I think the three doors that are really important are broader adopters from early adopters to people who are really evidence-based, and that includes KOLs and health systems and so forth. The second is FDA approval. FDA approval is really a stamp of an external party validating and really having reviewed the data. The third is coverage. Coverage is ultimately the end game when it comes to being able to offer a test that widely.
Another strong quarter recently, I think Galleri volumes grew 35% year-over-year. When you deconstruct that growth today, what are you focused on the most between adding new prescribers, driving utilization amongst physicians, and then setting up those digital health channels as well, which I think have been pretty important.
Right. When we think about who's ordering the test today, you have the individuals who are dabbling, so to speak, then you have high volume users, health systems. Our goal is really education is key. The dabblers, they're usually ones that patients come to them and say, "Oh, I've heard about this test. I want to take it," but they're not the ones who will actually drive it. Then there are the individuals who are high volume orders because they have actually seen the data, understand the data, then health systems which are at an organizational level are ready to adopt the test. All of that requires at the root of it an understanding of the test.
Our field team is out there educating physicians about the data, about the nuances of what makes a multi-cancer early detection test clinically useful, the benefits, and we put that together. The other element that's really important is making it easier for the test to be offered and to be deployed, so that's where a lot of our partnerships come into play.
A lot of data as well to discuss. What's feedback been since ASCO from the oncology and PCP community on the recent Galleri data, NHS, then the PATHFINDER 2 as well?
We were fortunate enough at ASCO to have two podium presentations. One was on our randomized control trial done in collaboration with NHS England. That study is called NHS-Galleri, and the other podium presentation was our U.S.-based multi-center prospective trial called PATHFINDER 2. NHS-Galleri had the headline news that we did not meet our pre-specified endpoint of a reduction in Stage III and IV. The nuances underneath that, and what we were able to present at ASCO was that we actually saw a reduction in Stage IV cancer. When the headline is we did not meet a reduction in combined Stage III and IV, underneath the hood is we actually did see a reduction in Stage IV. That was offset actually by an increase in Stage III.
This told us a lot about the biology, which is that there's actually a lot of Stage III cancer that is in people who don't suspect it, asymptomatic individuals, and that Galleri was able to detect them. If you look at the next layer of data, the reduction in Stage IV, the reduction in emergency room presentation, the fact that Galleri detected more Stage I and II cancers than all of the NHS guideline-recommended screening combined, that's really compelling data. Our field team has been taking that data and showing that to KOLs, to PCPs, and to health system executives, and other stakeholders to really show them that the next layer of the onion is really compelling data about the performance of the test.
The PATHFINDER 2 really is a great example of how this test performs very similarly in two different populations, and that level of validation is one that I think people who really understand the data come to appreciate.
Could you give us a little more context on the cancer types where you saw perhaps the best performance and maybe those that perhaps were lagging?
This is a conversation that goes on in the MCED field quite a bit, which is looking at aggregate performance over all the cancers in a population versus cancer by cancer. Obviously, we've been focused because it is multi-cancer early detection, because we detect a shared cancer signal across all the different cancer types. We've been focusing on that aggregate signal and shown that we have a very high positive predictive value north of 60%, which is 10x greater than single cancer screening. We show that our sensitivity for the 12 cancers that cause 2/3 of mortality, about 70% of the cancers we detected are in those cancer types and about 60% sensitivity. The other way people think about it is, well, maybe you should separate out cancers that we screen for, like breast, colorectal, and separate those from cancers that we don't screen for.
But in point of fact is we detected a large number of breast cancers and colorectal cancers and lung cancers because those are cancers that are not always detected by screening for a variety of reasons. People are not up to date with their screening. There are interval cancers. There are people who are survivors. More lung cancers are present in people who are not eligible for screening than who are eligible for screening, and those are ones that we can detect as well.
Very helpful. What do you think distinguishes a physician or practice that writes one to two Galleri orders from those who incorporate it routinely into their practice?
I mentioned to you the two kinds of phenotypes, the routine orderers and the dabblers, the one or two tests. What really distinguishes them is experience. Cancer is, although it's top of mind for everybody who works in the field, it's only in a 1%- 1.5% of people who are over 50. It's a relatively rare event. If someone only orders one or two tests, they're not likely to see that positive and get that positive feedback loop of, oh, I detected cancer and helped this individual. The physician who orders hundreds of tests will have that experience.
If you actually look at the public docket, and some physicians have actually written in prior to the advisory committee, which I'm sure we'll talk about, the physicians who've said, "I've ordered 500 or 600 or 1,000 tests, and I have seen patients who benefit," that's really the difference. The way to get the dabbler into the routine user is for them to understand the data and hopefully get that experience that tells them, oh, it is a valuable test.
The workflow improvements, I think have been pretty important there as well. The EHR integrations. Maybe just give a little color on what you've seen evolve there.
Ease of ordering is obviously reducing the friction. That's clearly a key step in adoption and increasing access. We have partnerships with, for example, Quest. We've had a good partnership with Quest for multiple years, renewed it last year, and there are two elements of that. One is integrating into the ordering system, and the other is actually integrating with Quest phlebotomy, so that the kit doesn't have to be mailed to a patient and then brought in. For example, they can just go to a Quest draw center, so that's very helpful. We also have integrations with ordering with athena and Epic. Again, anytime you can reduce friction, that helps with that ability to order and offer the test.
Want to spend some time now on the FDA AdCom coming up. The PMA, I think, is based upon the 12-month follow-up of the first 25,000 patients from PATHFINDER 2 and then the prevalent round of NHS-Galleri and the bridging study. What are the key factors, performance, and safety that you think the FDA is here to evaluate?
FDA approval is really predicated on a demonstration of safety and efficacy. They will ask a question on, do the probable benefits outweigh the probable risks? And those are questions that we address in both PATHFINDER 2 and NHS-Galleri. If you look at the performance characteristics, which is the effectiveness, we can demonstrate that we have very high positive predictive value, over 60%. We are able to, as I mentioned earlier, detect the really deadliest cancers and with high sensitivity. Close to 60% of our cancers are Stages I through III, and so that tells us that we are detecting early-stage cancers in asymptomatic individuals. If we look at the data from PATHFINDER 2, 80% of the individuals whose cancers were Stages I through III went on to receive curative intent therapy. So these are the cancers that are meaningful to detect in asymptomatic individuals.
On the safety side, we are able to show that we do not reduce adherence to guideline-recommended screening. It's really only a blood test, and so of course, routine blood tests are very safe. We show that there are very few invasive procedures. In fact, the rate of invasive procedures in a population that gets MCED is actually less than the rate of procedures in a population of women who get guideline-recommended mammography. So both from a safety and effectiveness standpoint, we have very strong data.
This isn't the first AdCom for GRAIL either. So maybe just talk us through the learnings from the 2023 one and what prep you're doing behind the scenes as well now ahead of next week.
In 2023, the FDA convened an AdCom on multi-cancer early detection. That's not quite three years ago. It was held in a November-ish timeframe three years ago. That was really to help educate the FDA from this advisory committee. They really had four takeaways that they said that are required of MCEDs. One is that performance characteristics. What are the performance characteristics of a high-quality MCED? The focus was on safety, and so very high positive predictive value, very high specificity. They said that really a benchmark should be 99.0% specificity. We easily cleared that benchmark with over 99.5% specificity in our validation data. A second characteristic is that physicians couldn't just be told, oh, we see a cancer signal, that they really needed guidance as to how to work up individuals.
This is encapsulated in something that is colloquially called a tissue of origin. We call it a Cancer Signal Origin for biological reasons, but we won't get into that. Essentially, this Cancer Signal Origin is a guide for the physician. You can imagine if we say, "Oh, the Cancer Signal Origin is lung," they would do a chest CT and so forth. We showed that we have over 90% accuracy in our Cancer Signal Origin. That was item number two. Item number three is that there should be clear labeling that an MCED test does not replace guideline-recommended screening. That is something that we have built into the safety profile of Galleri since the beginning. We are very clear about the role that MCED plays in the context of guideline-recommended screening.
As a consequence, we presented data earlier this year at AACR showing in the real world that women who got a negative Galleri result were still highly adherent to getting their mammograms on their regularly scheduled cadence. Fourth element may be a little bit more, we'll call it, controversial in the sense that the AdCom recommended randomized controlled trials. The randomized control trial is not necessary to demonstrate safety and effectiveness. It's much more useful in showing utility. As we pointed out for the AdCom next week and for FDA approval, it's focused on safety and efficacy. The 2023 advisory committee panels did say that to demonstrate utility, RCTs would be necessary. We've actually accomplished one in NHS-Galleri, and we've reported out on some utility measures, and other utility measures will still be forthcoming as we have longer follow-up.
It's very helpful. Just assuming the favorable decision there, what do you think changes commercially the day after that FDA approval? Are you anticipating a pretty noticeable effect on physician adoption, confidence, payer engagement? Maybe just lay out how you think that could perhaps evolve.
FDA approval really changes the landscape in the sense that it reduces some of the noise because it's very easy to point to this test has FDA approval and this one does not. I think that that clarity is going to be really important in helping physicians, patients, and payers really understand which test has undergone the rigorous validation and external review, because that's one of the most important things is the external review. That bifurcation is going to be really important as the landscape continues to change. The other element, too, is that it will open doors. Many payers will say, "You know what? We know that FDA approval is neither necessary nor sufficient, but it's really an important step along the way." Being able to say we have FDA approval is an important milestone that opens doors and increases payer engagement.
And we've had quite a few questions coming in on just the label and what that could look like. Talk us through your initial expectations for what that could perhaps be, and then how you would be preparing for different scenarios, how they could evolve.
Our go and position is because Galleri detects a cancer signal that's shared among all cancers, regardless of cancer types, that the most important characteristic of Galleri is the ability to report out a signal-detected result when we see it. The label itself is really about how we educate physicians and patients. There will be some who argue that the label should be restricted in the sense that one might say this test has been validated in cancers A, B, and C because there have been many examples of those. When we do a study in a population, the distribution of cancer types will be dependent on that population, and by definition, the most common cancers will appear more often than the rare cancers. It would not make sense for us to do a study that is sized for the most rare cancer possible.
It is conceivable that the labeling might state that there's relatively little data in the rare cancer type A, B, and C. But we believe that that's still okay, and in fact, it's scientifically accurate, and we will describe the distribution of cancers in our study. The other aspect of the labeling that will be very important will be how people should continue their guideline-recommended screening. At the end of the day, when we think about the potential outcomes on the labeling, I think we've been prepared for that because, A, we've been saying since the beginning of the Galleri launch that people should continue their guideline-recommended screening. And in all of our test reports and other documentation, we're very transparent about what cancers we've looked at and this data around each of the cancer types.
And how meaningful would differences in eligible age range or patient population be to that go-to-market strategy?
The labeling will be really dependent on the intended use population. In both NHS-Galleri and PATHFINDER 2, we enrolled individuals who are 50 and up. PATHFINDER 2 did not have an upper age limit, but NHS-Galleri had an upper limit of 77. Really, because age is the most important risk factor, one's cancer risk increases. If you look at it by age, it's almost like a hockey stick with starting at 50, it starts to accelerate. That age group is going to be really important. There have been some questions about, well, wouldn't this test be most beneficial in a, we'll call it, high-risk population as defined by, for example, genetic risk. But in fact, there's a difference between absolute risk and relative risk.
Like BRCA status, for example, increases your test in a relative fashion, but age is still the most important absolute risk factor. We believe that the data that we presented will really be focused on our intended use population as defined by age.
Then the final one specific to the AdCom , but are there any areas you are anticipating some pushback? Then how would GRAIL respond to those areas you're assuming there could be some frictions toward?
Yeah. We talked about aggregate versus cancer by cancer. We talked a little bit about that. The other area that gets debated in this field is the definition of early. If you ask what does the literature say about the definition of early, there's no single unified like this is the definition of early. Many people, therefore, like to revert to stage, and they'll say, "Oh, well, Stage I and II or Stage I, II and III is the definition for early." I think the fact that aggressive cancers shed cell-free nucleic acid, which is what we detect, is going to play a role in this. But to us, what really means early is are we detecting it before it would otherwise have been detected without Galleri?
We have to remember that our studies, PATHFINDER 2 and NHS-Galleri, were done in individuals who have no suspicion of cancer. By definition, that is early. What we can also show is that these individuals whose cancers we detect have a high likelihood of curative intent therapy. That translation and moving away strictly from stage is going to be an important part of the conversation. That notwithstanding, even if people say, "Well, I still want to define early by stage," we can show that 70% of our cancers that we detected were Stages I through III , and that should also be considered early.
Yeah, makes sense. Let us assume you get the FDA approval in the coming quarters. What is your expectation, then, for timing of CMS to open an NCD review?
We have been just having array conversations with CMS. As you know, there was the MCED Act that provides a pathway for CMS to cover MCED tests. Of course, CMS will still require their own evaluation and do their own analysis on certain data. We have been conversing with CMS on multiple interactions. It even goes back several years. We designed a study called REACH, which is in a Medicare beneficiary population. We designed that with them starting 2021, 2022. We are in the middle of enrolling that study. Really, our conversations with CMS will focus on the compendium of data that we have in a Medicare beneficiary population, not just in REACH, but in other data sources. Those conversations should accelerate after we get FDA approval.
Do you think you need the interim data from REACH before opening the NCD process?
No, I think that we do want to have those conversations, because the interim data for REACH will be important, but it is not the only data set. We have presented some compelling utility data at ASCO. We are going to continue to present that data over the course of the balance of this year as well as next year.
Makes sense. Post FDA, what do you think would need to occur for organizations like NCCN or the other guideline bodies to formally incorporate MCED testing into screening recommendations?
All of these guideline organizations, they will focus on really two aspects. One is, of course, the evidence, and the second is, of course, the KOLs who sit on those panels will need to weigh in and be favorably moved by the evidence. So number one is continue to generate that evidence. So we have strategies where shifting the narrative and conversation around validation, which is what FDA is really looking at, and to utility. We have published utility data, we will continue to publish utility data, and socializing that with the KOLs, getting them presented at major meetings. That will be really important. I think that on the payer side, the guidelines are important, but they are not the only drivers. So I think in parallel, we can have conversations with the KOLs who write guidelines.
After FDA approval, we will still be able to start having and accelerate conversations with payers themselves.
The team said commercial insurers are looking for that FDA approval first before those broader commercial coverage decisions come, which makes sense. And then if you think that box is checked, h ow do you expect those conversations to go with the commercial payers?
A point of fact is we actually already have some levels of coverage, either through employers, through self-insured employers, through TPAs, and so forth. We think that that could expand and accelerate some of those. Then I think with some of the commercial payers that, obviously that's the majority, but haven't initiated any sort of coverage, we can start having those conversations, maybe even do some pilots in select populations that are tailored to those specific payer populations and start generating data that is really fit for purpose.
When the team sits across the table from the leadership of a large health system, what do you think their biggest hurdle is today of that broader adoption?
I think it starts with evidence, then it becomes sort of deployment. The evidence base continues to grow. You've seen large health systems like Rush. They've recently announced their adoption of it and their offering of it, and that continues to increase. That is based on, obviously, the large body of clinical evidence. What's really gratifying is some health systems now are starting to publish their own experience. Mayo has published their experience, Dana-Farber has published their experience. Their experience is very much aligned with what we're seeing in our clinical studies, that high positive predictive value, that higher Cancer Signal Origin accuracy, patients who are benefiting. The more that gets out, that's going to encourage other health systems to really take a strong look at this.
The efforts that you mentioned earlier about making ease of ordering, whether that is Epic integration or work with athena and so forth, those will also be really important.
I want to cover competition a bit. We have more entrants coming into the market. From your perspective, that technical perspective, what do you think differentiates one MCED test from the other? I guess, talk about Galleri versus what else is out there.
As a scientist, the most important thing to me is the data, and there is no other MCED out there with the kind of data that we have. If we look at just the FDA package of NHS-Galleri of 140,000 enrolled, 70,000 in the interventional arm. If we look at PATHFINDER 2, 35,000 individuals, 25,000 of whom were in that FDA package. But in point of fact is we have a lot more data than that. We have real-world evidence data. We published a manuscript on 100,000 individuals that were, whose tests were run in the real world. We also have external data, and actually, that external data in some ways means more to the community than the GRAIL-generated data because that is their experience. I mentioned Dana-Farber, Mayo, and others are starting to publish as well.
I just actually saw something in the literature about a pancreatic patient who was detected, and that was published by a physician just on their own. I think these physician experiences, these health system experiences, are going to be really important. Data, I think, will be a key driver. The other thing, obviously, is FDA approval. FDA approval is, again, an example of an external review body rigorously looking through the data. Not only do they look at our data, but as you all know, they do inspections, and they have inspected us, they have inspected clinical trial sites. The rigor by which the FDA has evaluated the data is really a key differentiator.
We are evaluating the data on today's test, but how much better do you think Galleri could get over time with the R&D efforts that are perhaps on the way?
The Galleri that is offered today is not the first version. When we did PATHFINDER 1, that was on an earlier version, and we continue to evolve the test. I think there are several levels of evolution on the test that are really important. One is, obviously, performance characteristics. When I think about performance characteristics and what my future version of Galleri will be, and I am not saying next year, but as we iterate over time, I tell the younger people in the company, the Galleri that they take is going to be very different from the Galleri I take. When I think about it, we will be turning it into a movie. Right now, everything is a snapshot. It is based on a snapshot.
As we accumulate more data and get real-world evidence, as we see what happens to the signal over time, we can actually start making measurements and tailoring those measurements to each individual and turn that into a movie. I think that that is one element. The other element, I think, is going to be about throughput and cost. One of the key elements of GRAIL is that we have a beautiful facility in Research Triangle Park that can really scale this test, and that ability to scale will help us with things like COGS and throughput and so forth. There is that infrastructure that is going to be really important, and then there is going to be the science that is layered over top of that that really personalizes this test, and I think that is the direction we will be going.
Helpful. Want to spend just a few minutes on international. The Samsung partnership gives you access to South Korea, potentially Japan, and Singapore. What do you think Samsung brings that would be difficult for GRAIL to replicate itself? I guess, what could success in South Korea suggest about the best OUS model for Galleri?
So, it's early days with Samsung, but Samsung is obviously internationally known. It's got that name brand. It also has the experience of how to launch a product in the Korean market, and it also has an infrastructure that will be really important that we can learn from. So we're in the early days of that conversation, but we'll see what ways to leverage each of our respective capabilities effectively and efficiently. As you point out, it does have options for other Asian territories. I think, as you know, many countries have single-payer systems, and so we'll be able to understand how best to interact with these different single-payer systems. FDA approval will actually be really important. It's something that we can present to regulatory authorities in these different single-payer systems.
The other flip of the side of the coin, though, is that each health system is going to be different, and so we'll have to just sort of knock them out one by one.
Yeah, how have those discussions been with the payers? Obviously still early stages, but I think about the U.K. as well. My dad was actually part of the NHS-Galleri trial.
Oh, wow. Nice.
How have those early discussions been given the test has been out there and patients are somewhat familiar?
We're really appreciative of the collaboration we had with NHS England. I think that the most important part of our interaction has been the ability to really understand what implementation might look like. That implementation really goes all the way from the beginning to making people aware of the test. Then it goes to giving them the test, returning the result, and then in that fraction of individuals who get a signal, you just had to result triaging them to an efficient workup and then getting them to treatment earlier. The other element of implementation that we were able to do with NHS England was really about how do we get people to come back? That has been really an important part of the learning. There's a nice infrastructure there. Again, those conversations with NHS will continue.
Amazing. A closer from me. What do you think investors don't pay enough attention to from your discussions? I guess what is underappreciated the most about the GRAIL story?
Well, what's fascinating to me about GRAIL, as I mentioned, I've been at GRAIL for nine years, when I joined, it was really a hypothesis. In 2020, when we launched Galleri, it took about four years. In the five years since, we've improved on it, iterated on it. But we're really only at the beginning of the MCED curve. Not strictly speaking from an adoption standpoint, in terms of numbers, but in the number of patients who can benefit, that curve that we'll see in terms of adoption will impact the mortality curve, which we hope will see this. The other thing, too, is there's a really important interaction between detection, diagnosis, and treatment, all of us are in healthcare. If you go to ASCO, there was obviously the huge story about the RAS inhibitor.
Every ASCO, every ESMO, every AACR, there seems to be a new podium presentation about a new therapy that is going to be game changing. Our ability to detect these cancers early and get those individuals to these novel therapeutics earlier is really going to change that mortality curve. When I think about the benefit that Galleri has, I'm really thinking about it from the lens of we're making rapid advances in therapy, but we need to get people into that as early as possible, that's where a tool like Galleri can be really beneficial.
Just a little follow-up there, but across your nine years, you've obviously spent a lot of time doing many different things, but how is your time being spent right now, how does that perhaps compare to recent prior years? I guess a lot of it right now is ahead of next week.
Yes.
Maybe just speak a bit more broadly to that.
Yes. For me, I spend a lot of time thinking about two things. One is how do I effectively iterate on Galleri? I am working with the research team, I am working with the clinical team on a feedback loop between real-world data and the next generation Galleri version 4, version 5, version 6. That is where I spend a lot of my time. I actually spend a lot of time thinking about AI, and AI, of course, is infiltrating itself, for better or for worse, into everything we do. Galleri itself is a machine learning-based technology.
I do believe that as we think about these novel technologies around AI, I think that they can help us accelerate the advancement of my first goal, and ultimately the goal of everybody at GRAIL, which is to increase access to such a groundbreaking technology across populations, across countries, and really bend the mortality curve.
Amazing. Okay, Eric, thank you so much.
Well, thank you for having me. Really appreciate it.